delete_empty_descrs.php

<?php
// VERSION: 0.5
// delete empty info objs.
// In AtoM 2.4 is is possible to create empty Descriptions if a malformed
// csv is loaded. A common way to have this happen is to have blank lines
// present in the import CSV file.
//
// This script can be used to DELETE these empty Info Obj records.
//
// There is a $dryRun variable - set this to 'True' if you want to test
// what the sript finds WITHOUT DELETING. It will output a list of slugs
// it would have deleted. This is great for testing!
//
// The slugs created by this bug will look like 3 sets of 4 alphanum chars:
// e.g. n5ny-e7g3-w6te
//
// Setting $dryRun to 'False' will cause the script to delete any Descriptions
// it finds. MAKE BACKUPS FIRST!
//
// Setting $deleteChildren to 'True' will remove any child Descriptions found 
// attached to these empty descriptions. BE VERY CAREFUL WITH THIS SETTING. 
// TEST WITH DRY-RUN!!
//
// Command: php symfony tools:run <full_path_to_this_file>
//
// This script will:
// - Find all Descriptions that:
//   -- Are not the root info obj (ID is NOT 1)
//   -- That have a null (empty) identifier
//   -- That have a null title
//   -- That have a null repository ID
//   -- That have a null LOD value
//   -- Have the root info obj as a parent.
//
// - REBUILD ES INDEX AFTER RUNNING: php symfony search:populate

$dryRun = True;
$deleteChildDescriptions = False;

QubitSearch::disable();

function getChildren($id)
{
  $query = 'SELECT io.id, slug.slug
             FROM information_object io
             JOIN slug on slug.object_id = io.id
             WHERE io.parent_id = '.$id;

  $results = QubitPdo::fetchAll($query, array(), array('fetchMode' => PDO::FETCH_ASSOC));
  return $results;
}

$query = 'SELECT io.id, slug.slug, io.parent_id, io.identifier
            FROM information_object io
            JOIN information_object_i18n on io.id = information_object_i18n.id
            JOIN object ON io.id = object.id
            JOIN slug on slug.object_id = object.id
            WHERE object.id != '.QubitInformationObject::ROOT_ID.
          ' AND io.parent_id = 1
            AND information_object_i18n.title IS NULL
            AND io.identifier IS NULL
            AND io.level_of_description_id IS NULL
            AND io.repository_id IS NULL
            ORDER BY io.rgt ASC';

print("DELETING info obj records...\n");

$statement = QubitPdo::prepareAndExecute($query);
$counter = 0;

while ($row = $statement->fetch(PDO::FETCH_ASSOC))
{
  print ("Deleting information object. ID: " . $row['id'] . ' ' . $row['slug'] . ' ' . $row['identifier'] . "\n");

  try
  {
    $io = QubitInformationObject::getById($row['id']);
    
    $childList = array();
    
    if (null !== $children = getChildren($row['id']))
    {
      foreach ($children as $child)
      {
        $childList[] = $child["slug"];
      }
      
      if (0 < count($childList))
      {
        printf("Child recs found: %s\n", implode(",", $childList));
      }
    }
    
    if (!$dryRun && $deleteChildDescriptions && 0 < count($childList))
    {
      printf("Deleting child recs: %s\n", implode(",", $childList));
      $io->deleteFullHierarchy();
    }
    elseif (!$dryRun)
    {
      $io->delete();
    } 
    
    print(" DELETED.\n");
    $counter++;

  }
  catch (Exception $e)
  {
    print(" *ERROR - FAILED TO DELETE() OBJECT.\n");
    print($e);
  }
}
print("Deleted ".$counter." records.\n");