roblib_update_pmods/roblib_update_pmods.module

<?php

/**
 * @file
 * Module file for roblib_update_pmods.
 */

/**
 * Implements hook_cmodel_datastream_islandora_datastream_modified().
 */
function roblib_update_pmods_islandora_bookcmodel_mods_islandora_datastream_modified(AbstractObject $object, AbstractDatastream $datastream, array $params) {
  module_load_include('inc', 'islandora_paged_content', 'utilities');
  $pages = islandora_paged_content_get_pages($object);
  $parent_mods = $datastream->content;
  $pid = $datastream->parent->id;
  roblib_update_pmods_batch($parent_mods, $pages, $pid);
}

/**
 * Setup the batch.
 *
 * @param string $parent_mods
 *   The contents of the MODS datastream of the parent book.
 * @param Array $pages
 *   A list of pages related to the book
 * @param string $pid
 *   The PID of the book.
 */
function roblib_update_pmods_batch($parent_mods, $pages, $pid) {
  $batch = [
    'title' => t('Updating child page MODS records ...'),
    'operations' => [],
    'init_message' => t('starting'),
    'progress_message' => t('Processed @current out of @total.'),
    'error_message' => t('An error occurred during processing'),
    'finished' => 'roblib_update_pmods_batch_finished',
  ];

  foreach ($pages as $page) {
    /**
     * we use base64_encode to ensure we don't overload the batch
     * processor by stuffing complex objects into it
     */
    $batch['operations'][] = [
      '_roblib_update_pmods_update_child',
      [$page, $parent_mods],
    ];
  }
  batch_set($batch);
  batch_process('islandora/object/' . $pid);
}

/**
 * Handles individual requests from the batch.
 *
 * @param array $page
 *   A page from a book.
 * @param string $parent_mods
 *   The contents of the MODS datastream of the parent book.
 * @param array $context
 *   The batch context.
 */
function _roblib_update_pmods_update_child($page, $parent_mods, &$context) {
  if (!isset($context['results']['pages_processed'])) {
    $context['results']['pages_processed'] = 0;
  }
  $context['results']['pages_processed']++;
  $context['message'] = t('Processing MODS for page !p %pid', [
    '!p' => $context['results']['pages_processed'],
    '%pid' => $page['pid'],
  ]);
  try {
    roblib_update_pmods_update_child($page, $parent_mods, $context['results']['pages_processed']);
  } catch (Exception $e) {
    //TODO something
    watchdog('roblib_update_pmods', $e->getMessage(), NULL, WATCHDOG_ERROR, NULL);
  }
}

/**
 * Prepare a page object for update.
 *
 * @param array $page
 *   A page from a book.
 * @param string $parent_mods
 *   The contents of the MODS datastream of the parent book.
 * @param $image_number
 *   The image number of the book.
 *
 *
 * @throws \Exception
 */
function roblib_update_pmods_update_child($page, $parent_mods, $image_number) {
  $dsid = 'MODS';
  $mods_template = <<<EOT
  <mods xmlns="http://www.loc.gov/mods/v3" xmlns:mods="http://www.loc.gov/mods/v3" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsisnippet="http://www.w3.org/2001/XMLSchema-instance" version="3.2" xsisnippet:schemaLocation="http://www.loc.gov/mods/ http://www.loc.gov/standards/mods/mods.xsd">
  </mods>
EOT;
  $page_object = islandora_object_load($page['pid']);
  if (!isset($page_object[$dsid])) {
    $datastream = $page_object->constructDatastream($dsid);
    $datastream->label = 'MODS';
    $datastream->mimeType = 'application/xml';
    $datastream->content = $mods_template;
    // Ingest the datastream if it isn't yet.
  }
  else {
    $datastream = $page_object[$dsid];
  }
  roblib_update_pmods_compare_and_update($datastream, $parent_mods, $image_number);

  if (!isset($page_object[$dsid])) {
    $page_object->ingestDatastream($datastream);
  }
  unset($page_object);
}

/**
 * Remove XML nodes from a XML Domdocument.
 * @param array $nodes_to_delete
 *   The nodes to delete
 *
 * @throws \Exception
 */
function roblib_update_pmods_delete_nodes($nodes_to_delete){
  foreach($nodes_to_delete as $node) {
    $old_node = $node->parentNode->removeChild($node);
    if(empty($old_node)){
      throw new Exception ('Could not delete Node ' . $node->nodeValue);
    }
  }
}

/**
 * Update the page object MODS datastream if the parent MODS has been updated.
 * Only updates if the books accessCondition or the titleInfo/Title has been modified.
 *
 * @param AbstractDatastream $datastream
 *   A page objects MODS datastream
 * @param string $parent_mods
 *   A book objects MODS xml
 * @param $image_number
 *   The image number of the book.
 *
 * @throws \Exception
 */
function roblib_update_pmods_compare_and_update(&$datastream, $parent_mods, $image_number) {
  $updated = FALSE;
  $child_mods = $datastream->content;
  if (empty($child_mods)) {
    throw new Exception('Error loading Child MODS');
  }
  $child_xml_doc = new DOMDocument();
  $test = $child_xml_doc->loadXML($child_mods);
  if ($test === FALSE) {
    throw new Exception('Error processing Child MODS ' . $child_mods);
  }
  $child_xpath = new DOMXPath($child_xml_doc);
  $child_xpath->registerNamespace('mods', 'http://www.loc.gov/mods/v3');
  $child_xpath->registerNamespace('xlink', 'http://www.w3.org/1999/xlink');
  $query = "/mods:mods/mods:titleInfo[not(@type)]/mods:title";
  $parent_xml_doc = new DOMDocument();
  $test = $parent_xml_doc->loadXML($parent_mods);
  if ($test === FALSE) {
    throw new Exception('Error processing Parent MODS ' . $parent_mods);
  }
  $parent_xpath = new DOMXPath($parent_xml_doc);
  $parent_xpath->registerNamespace('mods', 'http://www.loc.gov/mods/v3');
  $parent_xpath->registerNamespace('xlink', 'http://www.w3.org/1999/xlink');
  $parent_results = $parent_xpath->query($query, $parent_xml_doc);
  $child_results = $child_xpath->query($query, $child_xml_doc);
  $p_arr = [];
  $parent_book_title_arr = [];
  $c_arr = [];
  $nodes_to_delete = [];
  $image_number_string = 'Image ' . $image_number . ' - ';
  foreach ($child_results as $child_node) {
    $c_arr[] = $child_node->nodeValue;
    $nodes_to_delete[] = $child_node->parentNode;
  }
  foreach ($parent_results as $p_node) {
    $p_arr[] = $image_number_string . trim($p_node->nodeValue);
    $parent_book_title_arr[] = trim($p_node->nodeValue);
  }
  $missing_titles = array_diff($p_arr, $c_arr);

  if(count($missing_titles) > 0) {
    $updated = TRUE;
    roblib_update_pmods_delete_nodes($nodes_to_delete);
  }
  foreach ($missing_titles as $title) {
    roblib_update_pmods_update_title($child_xml_doc, $title);
  }
  $book_title_query = "/mods:mods/mods:relatedItem[@type='host']/mods:titleInfo/mods:title";
  $child_book_title_results = $child_xpath->query($book_title_query, $child_xml_doc);
  $c_arr = array();
  $nodes_to_delete = [];
  foreach($child_book_title_results as $c_node) {
    $c_arr[] = $c_node->nodeValue;
    $nodes_to_delete[] = $c_node->parentNode->parentNode;
  }
  $missing_titles = array_diff($parent_book_title_arr, $c_arr);
  if(count($missing_titles) > 0) {
    $updated = TRUE;
    roblib_update_pmods_delete_nodes($nodes_to_delete);
  }
  foreach ($missing_titles as $title) {
    roblib_update_pmods_update_book_title($child_xml_doc, $title);
  }
  $access_condition_query = "/mods:mods/mods:accessCondition[@type='use and reproduction' and @xlink:href='http://rightsstatements.org/vocab/InC/1.0/']";
  $child_access_condition_results = $child_xpath->query($access_condition_query, $child_xml_doc);
  $parent_access_condition_results = $parent_xpath->query($access_condition_query, $parent_xml_doc);
  $p_arr = array();
  $c_arr = array();
  $nodes_to_delete = [];
  foreach($child_access_condition_results as $c_node) {
    $c_arr[] = $c_node->nodeValue;
    $nodes_to_delete[] = $c_node;
  }
  foreach ($parent_access_condition_results as $p_node) {
    $p_arr[] = $p_node->nodeValue;
  }
  $missing_conditions = array_diff($p_arr, $c_arr);
  if(count($missing_conditions) > 0 || $parent_access_condition_results->length < 1) {
    $updated = TRUE;
    roblib_update_pmods_delete_nodes($nodes_to_delete);
  }
  foreach ($missing_conditions as $condition) {
    roblib_update_pmods_update_conditions($child_xml_doc, $condition, TRUE);
  }
  // A book may have a custom accessCondition that doesn't link to a license
  $access_condition_query = "/mods:mods/mods:accessCondition[@type='use and reproduction' and not(@xlink:href)]";
  // TODO refactor duplicate code into function
  $child_access_condition_results = $child_xpath->query($access_condition_query, $child_xml_doc);
  $parent_access_condition_results = $parent_xpath->query($access_condition_query, $parent_xml_doc);
  $p_arr = array();
  $c_arr = array();
  $nodes_to_delete = [];
  foreach($child_access_condition_results as $c_node) {
    $c_arr[] = $c_node->nodeValue;
    $nodes_to_delete[] = $c_node;
  }

  foreach ($parent_access_condition_results as $p_node) {
    $p_arr[] = $p_node->nodeValue;
  }
  $missing_conditions = array_diff($p_arr, $c_arr);
  if(count($missing_conditions) > 0 || $parent_access_condition_results->length < 1) {
    $updated = TRUE;
    roblib_update_pmods_delete_nodes($nodes_to_delete);
  }
  foreach ($missing_conditions as $condition) {
    roblib_update_pmods_update_conditions($child_xml_doc, $condition, FALSE);
  }
  if ($updated) {
    $updated_xml = $child_xml_doc->saveXML();
    $datastream->setContentFromString($updated_xml);
  }

}

/**
 * Update the accessCondition node of the pages MODS xml.
 *
 * @param \DOMDocument $mods_doc
 *   A pages MODS xml
 * @param string $condition
 *   The new accessCondition nodes value.
 * @param bool $use_xlink
 *   If true it will add the xlink:href attribute.
 */
function roblib_update_pmods_update_conditions($mods_doc, $condition, $use_xlink = FALSE){
  $root = $mods_doc->documentElement;
  $accessCondition = $mods_doc->createElement('accessCondition', $condition);
  $root->appendChild($accessCondition);
  $type_attribute = $mods_doc->createAttribute('type');
  $type_attribute->value = 'use and reproduction';
  $accessCondition->appendChild($type_attribute);
  if($use_xlink){
    $href_attribute = $mods_doc->createAttribute('xlink:href');
    $href_attribute->value = 'http://rightsstatements.org/vocab/InC/1.0/';
    $accessCondition->appendChild($href_attribute);
  }
}

/**
 * Update the page MODS xml with a relatedItem[@type='host'] title with the books title.
 *
 * @param \DOMDocument $mods_doc
 *   A page MODS xml
 * @param string $book_title
 *   The title of the book.
 */
function roblib_update_pmods_update_book_title (&$mods_doc, $book_title) {
  $root = $mods_doc->documentElement;
  $related_item = $mods_doc->createElement('relatedItem');
  $type_attribute = $mods_doc->createAttribute('type');
  $type_attribute->value = 'host';
  $related_item->appendChild($type_attribute);
  $root->appendChild($related_item);
  $title_info = $mods_doc->createElement('titleInfo');
  $related_item->appendChild($title_info);
  $title = $mods_doc->createElement('title', (string) $book_title);
  $title_info->appendChild($title);
}

/**
 * Update the page MODS xml titleInfo/title with the books title prefixed by image #.
 *
 * @param \DOMDocument $mods_doc
 *   A page MODS xml
 * @param string $book_title
 *   The title of the book prefixed with the page image #.
 */
function roblib_update_pmods_update_title(&$mods_doc, $title) {
  $root = $mods_doc->documentElement;
  $title_info = $mods_doc->createElement('titleInfo');
  $root->appendChild($title_info);
  $title = $mods_doc->createElement('title', (string) $title);
  $title_info->appendChild($title);
}

/**
 * @param $success
 * @param $results
 * @param $operations
 *
 * @return array|mixed|string|null
 */
function roblib_update_pmods_batch_finished($success, $results, $operations) {
  return t('Finished Processing Pages.');
}