idu.module

<?php

/**
 * Implements hook_init.
 * 
 * @return type
 */
function idu_init( ) {
  // module_load_include('inc', 'idu', 'idu_constants');
  // module_load_include('inc', 'repository_control');
  // module_load_include('inc', 'repository_control', 'log_record');
  // module_load_include('inc', 'repository_control', 'csv_cells');
  // module_load_include('inc', 'repository_control', 'object_record');
  // module_load_include('inc', 'phpexcel');
  if (module_exists('islandora_paged_content')) {
    module_load_include('inc', 'islandora_paged_content', 'includes/utilities');
    module_load_include('inc', 'islandora_paged_content', 'includes/manage_pages');
  }
  if (module_exists('islandora_audio')) { module_load_include('inc', 'islandora_audio', 'includes/derivatives'); }
  if (module_exists('islandora_fits')) { module_load_include('inc', 'islandora_fits', 'includes/derivatives'); }
  if (module_exists('islandora_basic_image')) { module_load_include('inc', 'islandora_basic_image', 'includes/derivatives'); }
  if (module_exists('islandora_large_image')) { module_load_include('inc', 'islandora_large_image', 'includes/derivatives'); }
  if (module_exists('islandora_pdf')) { module_load_include('inc', 'islandora_pdf', 'includes/derivatives'); }
  return;
}

/** 
 * Implements hook_help.
 * 
 * @param type $path
 * @param type $arg
 * @return type
 */
function idu_help($path, $arg) {
  if ($path == 'admin/help#idu') {
    return t('Islandora Drush Utilities provides a suite of commands and options to help admins manage Fedora objects and their datastreams.');
  }
}


/**
 * Implements hook form.
 * 
 * @param type $form
 * @param type $form_state
 * @return type
 */
function idu_main_form($form, &$form_state) {
  if (!isset($form_state['storage']['confirm'])) {
    
    $form['operation'] = array(
        '#type' => 'select',
        '#title' => t('Select the operation to be performed.'),
        '#default_value' => variable_get('icg_operation', 'HEALTH_CHECK'),
        '#options' => array(
          'OPEN_REPORT' => t('  OPEN_REPORT - Present a link to download / open the specified XLSX health check report. '),
        ),
    );
    
   $form['submit'] = array(
        '#type' => 'submit',
        '#value' => t('Submit'),
    );

    return $form;
    
  } else {  // The confirmation (secondary) form.
    
    $op = $form_state['values']['operation'];
    $default_file = variable_get('icg_xlsxFile', "public://HEALTH_CHECK.xlsx");
    
    if ($op === 'OPEN_REPORT') {
      $form['xlsxFile'] = array(
       '#type' => 'textfield',
       '#title' => t('Enter the name of the existing XLSX file to open. '),
       '#default_value' => $default_file,
       '#size' => 80,
       '#maxlength' => 80,
      );
    } else {
      $form['first'] = array(
       '#type' => 'textfield',
       '#title' => t('Enter the first PID number to process.'),
       '#default_value' => variable_get('icg_first', '0'),
       '#size' => 10,
       '#maxlength' => 10,
      );
      $form['last'] = array(
        '#type' => 'textfield',
        '#title' => t('Enter the last PID number to process.'),
        '#default_value' => variable_get('icg_last', '10000'),
        '#size' => 10,
        '#maxlength' => 10,
      );
      $form['inactives'] = array(
        '#type' => 'checkbox',
        '#title' => t('Check here if you want to include inactive and empty (deleted) objects.'),
        '#default_value' => variable_get('icg_inactives', FALSE),
      );
    }

  }
    
  $alias = variable_get('icg_repoAlias');
  $namespace = variable_get('icg_namespace');
  return confirm_form($form, " Operation '$op' - Proceed with the specified inputs and the '$namespace' namespace of '$alias'? ", 'idu', '  This action cannot be undone. ', $op, 'Cancel');
  
}

/**
 * Implements hook_form_validate.
 *  
 * @param type $form
 * @param type $form_state
 * @return type
 */
function idu_main_form_validate($form, &$form_state) {
  return;
}

/**
 * Implements hook_form_submit.
 * 
 * @param type $form
 * @param type $form_state
 * @return type
 */
function idu_main_form_submit($form, &$form_state) {

  // If confirm is NOT set, this will cause the form to be rebuilt, entering the confirm part.
  if (!isset($form_state['storage']['confirm'])) {

    $form_state['storage']['confirm'] = TRUE;
    $form_state['storage']['original_form'] = $form_state['values'];
    $form_state['rebuild'] = TRUE;
      
  } else {  // Do real form submittal.
    
    drupal_set_message(t('The Islandora Health Check main form has been submitted.'));

    $main = $form_state['storage']['original_form'];   // get controls from the main form
    $secondary = $form_state['values'];                // get controls from the secondary form

    icg_set_variable('operation', $main);

    icg_set_variable('first', $secondary);
    icg_set_variable('last', $secondary);
    icg_set_variable('inactives', $secondary);
    icg_set_variable('onlyCols', $secondary);
    icg_set_variable('xlsxFile', $secondary);

    idu_op( );  // Distribute control to the specified $op.
    }
    
  return;
}

/**
 * Implements hook_menu().
 * 
 * @return type
 */
function idu_menu( ) {
  $items = array();
  $items['idu'] = array(
      'title' => 'Islandora Drush Utilities',
      'page callback' => 'drupal_get_form',
      'page arguments' => array('idu_main_form'),
      'description' => 'Islandora Drush Utilities',
      'access arguments' => array('administer users'),
      'type' => MENU_NORMAL_ITEM,
  );
  return $items;
}

/**
 * Returns an associative array of PIDS found in the specified range.
 * 
 * @param array $context
 *   Drupal batch $context array.  Returns with an associative array of 
 *   PIDnum => Title values in $context['sandbox']['objPIDs'] for objects 
 *   found in the specified range.
 * @return integer
 *   The PID number of the last/highest numbered PID found in the range. The
 *   variable icg_last is automatically set to this limiting value.
 */
function idu_getPIDs(&$context) {
  $obj_pids = array();
  $first = intval(variable_get('icg_first'));
  $last = intval(variable_get('icg_last'));
  $range = $last - $first + 1;
  $largest = 0;
  
  $ns = variable_get('icg_namespace');

  // Loop through the repo looking for objects in the specified range.

  for ($i=$first, $notFound=0; $i<=$last; $i++) {
    $pid = $ns.$i;
    try {
      $object = idu_repo( )->getObject($pid);
    } catch (Exception $e) {
      if ($e->getCode() === 404) {
        $obj_pids[$i] = '-*-Not Found-*-';
        $notFound++;
      } else {
        drupal_set_message("PID '$pid' forced a FedoraRepository getObject( ) exception: '$e' ", 'error');
      }
      continue;  // for any exception...skip this PID
    }
    $obj_pids[$i] = $object->label;
    $largest = $i;
  }

  // If any PIDs were not found, report them.
  if ($notFound > 0) {
    if ($range > 10) {
      drupal_set_message("A total of $notFound PIDs in the specified range were NOT FOUND.", 'warning');
    }
    foreach ($obj_pids as $pid => $status) {
      if ($status === '-*-Not Found-*-') {
        if ($range<=10 && variable_get('icg_verbose')) {
          drupal_set_message("Object $pid was NOT FOUND.", 'warning');
        }
        unset($obj_pids[$pid]);  // remove it from our list
      }
    }
  }

  // Save the found object PIDs in the batch results.
  $context['results']['objectPIDs'] = $obj_pids;
  idu_post($context);
  
  variable_set('icg_last', $largest);
  return $largest;
}

/**
 * Open the repository for the rest of IDU to use.
 * 
 * @global FedoraRepository $repository
 * @return FedoraRepository 
 */
 function idu_repo( ) {
  global $repository;
  if (!is_null($repository)) {
    return $repository;
  } else if (!$repository = repository_control_connect()) { 
    die('Error in idu_repo( )...no connection to Fedora!'); 
  } 
  return $repository;
}


/**
 * Create or Modify an object handle and assign it.
 *
 * This function relies on constants defined in idu_constants.inc.  It attempts to open a new or existing Handle system
 * batch command file and appends necessary commands to this file to create or modify and assign a new handle to the target
 * object.  The batch file is defined in temporary://H_BATCH.  This function does NOT initiate processing of the batch file,
 * that must be done elsewhere.
 *
 * @param FedoraObject $object
 *   The Fedora object for which the handle is created or modified.
 * @param string $op
 *   Set to either 'CREATE' or 'MODIFY'.  If --force is specified this value may be overridden.
 * @param bool $force
 *   If TRUE, this function will continue even if the state of the mods:identifier[@type='hdl'] element is not as expected for the specified $op.
 * @return int
 *   Returns 0 or 1 indicating the number of handles processed.
 */
function idu_handle_create_or_modify($object, $op='CREATE', $force=FALSE) {

  $pid = $object->id;

  // Specify the target batch and log files in temporary://.
  $fileName = 'temporary://'.H_BATCH;
  $logFile = 'temporary://'.H_LOG;

  // Open (or create) the batch file and log file.
  $old = file_exists($fileName);
  if (!$file = fopen($fileName, 'a+')) die("Could not open batch file $fileName!  ");
  if (!$log = fopen($logFile, 'a+')) die("Could not open log file $logFile!  ");

  // Open the MODS datastream, if none... done.
  if (!$mods = $object['MODS']) {
    fprintf($log, "Warning: Object '$pid' has NO MODS record.\n");
    fclose($file);
    fclose($log);
    return 0;
  }

  // Make the MODS content searchable as a SimpleXMLElement object.
  $content = $mods->content;
  $xml = new SimpleXMLElement($content);
  $xpath = "/mods:mods/mods:identifier[@type='hdl']";

  // Does this object already have a handle?
  if ($nodes = $xml->xpath($xpath)) {    // Yes...
    $oldHandle = (string)$nodes[0];
    if ($op === 'CREATE' && !$force) {
      fclose($file);
      fclose($log);
      return 0;
    }
  } else if ($op === 'MODIFY') {        // No existing handle...if Modify we need to adjust or give up.
    if ($force) {
      $op = 'CREATE';
    } else {
      fprintf($log, "Warning: Object '$pid' has no assigned Handle.  Use --force or 'drush iduH $pid Create' if you wish to create one.\n\n");
      fclose($file);
      fclose($log);
      return 0;
    }
  }

  // Appropriate handle status found.  Create or modify one as needed.
  list($ns, $pidn) = explode(':', $pid);

  // If this is a new batch command file, print an AUTHENTICATE command block.
  if (!$old) {
    $cmd  = "AUTHENTICATE PUBKEY:300:0.NA/".H_PREFIX." \n".H_SVRDIR."/admpriv.bin";
    if (H_ADMKEY == ' ') {
      $key = $keyLog = '';
    } else {
      $key = '|'.H_ADMKEY;
      $keyLog = '|***obfuscated*keyphrase***';
    }
    fprintf($file, "%s%s\n\n", $cmd, $key);
    fprintf($log, "%s%s\n\n", $cmd, $keyLog);
  }

  // Set the batch file permissions so that it can be deleted later.
  $path = drupal_realpath($fileName);
  if (!$ok = chmod($path, 0777)) {
    die("Permissions could not be set for batch file $fileName.  ");
  }

  // Build the object handle and handle.net URI for this object.  If the namespace of the object matches the specified
  //   H_NUMERIC_ONLY defined parameter, then use only the numeric part of the $pid; otherwise use the entire $pid.
  //
  $handle = ("$ns:" === H_NUMERIC_ONLY ? $pidn : $pid);
  $hdl = 'http://hdl.handle.net/'.H_PREFIX.'/'.$handle;

  // Sanitize the object ID (title) for Handle!  No double quotes and length no more than 120 characters.
  $handle_title = str_replace('"', "", $object->label);
  if (strlen($handle_title) > 117) {
    $lines = explode("\n", wordwrap($handle_title, 117));
    $handle_title = $lines[0] . "...";
  }

  // Remove all old handle identifiers and add one new HDL identifier element to the MODS.
  $mods = new DOMDocument( );
  $mods->loadXML($content);
  $xpath = new DOMXpath($mods);

  foreach ($xpath->query("/mods:mods/mods:identifier[@type='hdl']") as $node) {
    $parent = $node->parentNode;
    $parent->removeChild($node);    // remove the old handles
  }

  // Add a new type='hdl' identifier.
  $id = $mods->createElement('identifier', $hdl);
  $mods->documentElement->appendChild($id);
  $id->setAttribute('type', 'hdl');

  // Remove all URI and DOI identifier elements found at the first level.
  $types = array("doi", "uri");
  foreach ($types as $type) {
    foreach ($xpath->query("/mods:mods/mods:identifier[@type='$type']") as $node) {
      $parent = $node->parentNode;
      $parent->removeChild($node);    // remove the old element
    }
  }

  // If the object has a 'local' identifier, preserve it.  If not, add the handle as a new local ID.
  $nLocal = 0;
  foreach ($xpath->query("/mods:mods/mods:identifier[@type='local']") as $node) {
    $nLocal++;
  }

  if ($nLocal === 0) {
    $id = $mods->createElement('identifier', $pid);
    $mods->documentElement->appendChild($id);
    $id->setAttribute('type', 'local');
  }

  /* Do NOT add a new type='uri' identifier to match the handle ID.
  $id = $mods->createElement('identifier', $hdl);
  $mods->documentElement->appendChild($id);
  $id->setAttribute('type', 'uri');
  */

  // Save the MODS changes.
  $object['MODS']->setContentFromString($mods->saveXML());

  // Add the appropriate command to the batch file.
  $cmd = "$op ".H_PREFIX."/$handle";
  fprintf($file, "%s\n", $cmd);
  fprintf($log, "%s\n", $cmd);

  $cmd  = "100 HS_ADMIN 86400 1110 ADMIN 300:111111111111:0.NA/".H_PREFIX."\n";
  $cmd .= "8 URL 86400 1110 UTF8 ".H_OBJECT_PATH."$pid\n";
  $cmd .= "9 DESC 86400 1110 UTF8 $handle_title\n";
  fprintf($file, "%s\n", $cmd);
  fprintf($log, "%s\n", $cmd);

  // Close the batch and log files and return 1, the number of objects processed.
  fclose($file);
  fclose($log);

  return 1;
}