islandora_find_replace.module

<?php

/**
 * @file
 * Main module file for Islandora Find & Replace.
 */

/**
 * Implements hook_menu().
 */
function islandora_find_replace_menu() {
  $items = array();
  $items['admin/islandora/tools/find-replace/find'] = array(
    'title' => 'Find & Replace',
    'description' => 'Search for instances of text.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('islandora_find_replace_find_form'),
    'access arguments' => array('use islandora find & replace'),
    'file' => 'includes/find.form.inc',
    'type' => MENU_NORMAL_ITEM,
  );
  $items['admin/islandora/tools/find-replace/log'] = array(
    'title' => 'Find & Replace Logs',
    'description' => 'Log of Find & Replace datastream modifications',
    'page callback' => 'islandora_find_replace_operation_logs',
    'access arguments' => array('use islandora find & replace'),
    'file' => 'includes/logs.inc',
    'type' => MENU_NORMAL_ITEM,
  );
  $items['admin/islandora/tools/find-replace/log/get_csv'] = array(
    'title' => 'Generate CSV Log',
    'description' => 'Gererate CSV Log file',
    'page callback' => 'generate_csv',
    'access arguments' => array('use islandora find & replace'),
    'file' => 'includes/logs.inc',
    'type' => MENU_LOCAL_ACTION,
  );
  $items['admin/islandora/tools/find-replace/replace/%islandora_find_replace'] = array(
    'title' => 'Find & Replace Text',
    'description' => 'Replace instances of text.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('islandora_find_replace_replace_form', 5),
    'access arguments' => array('use islandora find & replace'),
    'file' => 'includes/replace.form.inc',
    'type' => MENU_CALLBACK,
  );
  $items['admin/islandora/tools/find-replace/log/%islandora_find_replace'] = array(
    'title' => 'Find & Replace Operation Log',
    'description' => 'Log of Find & Replace datastream modifications',
    'page callback' => 'islandora_find_replace_operation_log',
    'page arguments' => array(5),
    'access arguments' => array('use islandora find & replace'),
    'file' => 'includes/log.inc',
    'type' => MENU_CALLBACK,
  );
  $items['admin/islandora/tools/find-replace/replace/%islandora_find_replace/%islandora_object/preview'] = array(
    'title' => 'Preview Change',
    'description' => 'Preview how the datastream will change.',
    'page callback' => 'islandora_find_replace_preview',
    'page arguments' => array(5, 6),
    'access callback' => 'islandora_find_replace_preview_access',
    'access arguments' => array(5, 6),
    'type' => MENU_CALLBACK,
  );
  return $items;
}

/**
 * Loads the data for a find & replace operation.
 *
 * @param int $id
 *   The record's ID.
 *
 * @return array
 *   The database row as an array.
 */
function islandora_find_replace_load($id) {
  return db_query("SELECT * FROM {islandora_find_replace} WHERE id = :id LIMIT 1",
    array(':id' => $id))->fetchAssoc();
}

/**
 * Implements hook_permission().
 */
function islandora_find_replace_permission() {
  return array(
    'use islandora find & replace' => array(
      'title' => t('Use the Islandora Find & Replace form'),
      'description' => t('Users will also need permission to modify each object. Users will be able to view all Find & Replace logs.'),
    ),
  );
}

/**
 * Access callback for the datastream diff preview.
 *
 * @param array $find_replace
 *   The find & replace operation data.
 * @param FedoraObject $object
 *   The Fedora object.
 *
 * @return bool
 *   TRUE if the user has access.
 */
function islandora_find_replace_preview_access($find_replace, $object) {
  if (!module_exists('islandora_pretty_text_diff')) {
    return FALSE;
  }

  if (!user_access('use islandora find & replace')) {
    return FALSE;
  }

  if (user_access('view datastream diff') && islandora_datastream_access(ISLANDORA_VIEW_DATASTREAM_HISTORY, $object[$find_replace['dsid']])) {
    return TRUE;
  }
  return FALSE;
}

/**
 * Make a Unix timestamp for the start or end of a day.
 *
 * @param array $date
 *   From the Drupal date form element value.
 * @param bool $start
 *   TRUE if requesting the start of the day.
 *
 * @return int
 *   The Unix timestamp.
 */
function _islandora_find_replace_timestamp_from_form_date($date, $start = TRUE) {
  if ($start) {
    return gmmktime(0, 0, 0, $date['month'], $date['day'], $date['year']);
  }
  else {
    return gmmktime(23, 59, 59, $date['month'], $date['day'], $date['year']);
  }
}

/**
 * Batch operation to filter potential objects to perform the text replacement.
 *
 * @param array $candidate
 *   Potential object found with the SPARQL query.
 * @param string $dsid
 *   The datastream ID for the operation.
 * @param string $search
 *   The search ("find") string for the operation.
 * @param int $submission_id
 *   The database ID for this submission/operation.
 * @param array $context
 *   Batch context.
 */
function islandora_find_replace_get_targets($candidate, $dsid, $search, $submission_id, &$context) {
  if (!isset($context['results']['submission_id'])) {
    $context['results']['submission_id'] = $submission_id;
  }
  $object = islandora_object_load($candidate['object']);

  if (isset($object[$dsid]) && $count = substr_count($object[$dsid]->content, $search)) {
    if (islandora_datastream_access(ISLANDORA_METADATA_EDIT, $object[$dsid])) {
      $candidate['count'] = $count;
      $context['results']['find_results'][] = $candidate;
    }
  }
}

/**
 * The completion callback for the "find" batch.
 *
 * @param bool $success
 *   If the batch successfully completed.
 * @param array $results
 *   An array of result data.
 * @param array $operations
 *   Unprocessed operations if $success is FALSE.
 */
function islandora_find_replace_find_complete($success, $results, $operations) {
  $find_replace = islandora_find_replace_load($results['submission_id']);
  if (isset($results['find_results'])) {
    $find_results = $results['find_results'];
  }
  else {
    $find_results = array();
  }
  $find_replace['find_results'] = $find_results;
  $find_replace['state'] = 'search';
  drupal_write_record('islandora_find_replace', $find_replace, 'id');
}

/**
 * Batch operation to perform find & replace on an object.
 *
 * @param string $pid
 *   The object's PID.
 * @param string $dsid
 *   The datastream ID to operate on.
 * @param string $search
 *   The search string that will be replaced.
 * @param string $replace
 *   The replacement string.
 * @param int $id
 *   The find & replace submission/operation ID.
 * @param array $context
 *   The batch context.
 */
function islandora_find_replace_update_objects($pid, $dsid, $search, $replace, $id, &$context) {
  if (!isset($context['results']['submission_id'])) {
    $context['results']['submission_id'] = $id;
  }
  $object = islandora_object_load($pid);
  $success = FALSE;
  if (isset($object[$dsid]) && islandora_datastream_access(ISLANDORA_METADATA_EDIT, $object[$dsid])) {
    try {
      $location_before = $object[$dsid]->location;
      $content = islandora_find_replace_string_replace($search, $replace, $object[$dsid]->content);
      $object[$dsid]->setContentFromString($content);

      $location_after = $object[$dsid]->location;
      // If successful, track the version.
      if ($location_before != $location_after) {
        $success = TRUE;
      }
    }
    catch (Exception $e) {

    }
  }

  $find_replace = islandora_find_replace_load($id);
  $revisions = unserialize($find_replace['revisions']);

  if ($success) {
    if (!isset($revisions['success'])) {
      $revisions['success'] = array();
    }
    $revisions['success'][$pid] = array('before' => $location_before, 'after' => $location_after);
  }
  else {
    if (!isset($revisions['fail'])) {
      $revisions['fail'] = array();
    }
    $revisions['fail'][] = $pid;
  }
  $find_replace['revisions'] = $revisions;
  drupal_write_record('islandora_find_replace', $find_replace, 'id');
}

/**
 * The completion callback for the "replace" batch.
 *
 * @param bool $success
 *   If the batch successfully completed.
 * @param array $results
 *   An array of result data.
 * @param array $operations
 *   Unprocessed operations if $success is FALSE.
 */
function islandora_find_replace_update_complete($success, $results, $operations) {
  
      
  db_update('islandora_find_replace')
    ->fields(array('state' => 'complete'))
    ->condition('id', $results['submission_id'])
    ->execute();
  drupal_set_message(t('Find and Replace complete!'));
}

/**
 * Peform a string replacement.
 *
 * @param string $search
 *   The search/find string.
 * @param string $replace
 *   The replacement string.
 * @param string $content
 *   The target string for the operation.
 *
 * @return mixed
 *   The modified string if successful.
 */
function islandora_find_replace_string_replace($search, $replace, $content) {
  if ($replace == '(empty)') {
    $replace = '';
  }
  return str_replace($search, $replace, $content);
}

/**
 * Find objects to perform the find/replace operation based on user form values.
 *
 * @param string $model
 *   The content model PID that the operation should apply to.
 * @param bool|string $collection
 *   The collection PID if the user selects a collection.
 * @param bool|string $date_property
 *   The type of date filter to apply if the user has selected.
 * @param bool|string $date_from
 *   The start date for the date filter if selected.
 * @param bool|string $date_to
 *   The end date for the date filter if selected.
 *
 * @return array
 *   An array of SPARQL results.
 */
function islandora_find_replace_query($model, $collection = FALSE, $date_property = FALSE, $date_from = FALSE, $date_to = FALSE) {
  $condition = '?object fm:hasModel <info:fedora/' . $model . '> ; ';

  if ($collection) {
    $condition .= 'fr:isMemberOfCollection <info:fedora/' . $collection . '> ; ';
  }
  if ($date_property) {
    $from = _islandora_find_replace_timestamp_from_form_date($date_from);
    $to = _islandora_find_replace_timestamp_from_form_date($date_to, FALSE);
    $from = gmdate("Y-m-d\TH:i:s\Z", $from);
    $to = gmdate("Y-m-d\TH:i:s\Z", $to);

    $date_query = ($date_property == 'lastModifiedDate') ? 'fv:lastModifiedDate' : 'fm:createdDate';
    $condition .= $date_query . ' ?date
      FILTER ( ?date >= xsd:dateTime("' . $from . '") && ?date <= xsd:dateTime("' . $to . '"))';
  }
  $tuque = islandora_get_tuque_connection();
  $query = "PREFIX fm: <" . FEDORA_MODEL_URI . ">
            PREFIX fr: <" . FEDORA_RELS_EXT_URI . ">
            PREFIX fv: <info:fedora/fedora-system:def/view#>
            PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
            SELECT ?object ?label
            FROM <#ri>
            WHERE {
              " . $condition . "
              OPTIONAL{
                ?object fm:label ?label
              }
           }";
  $results = array();
  $query_results = $tuque->repository->ri->sparqlQuery($query, 'unlimited');
  foreach ($query_results as $result) {
    $results[] = array('label' => $result['label']['value'], 'object' => $result['object']['value']);
  }
  return $results;
}

/**
 * Page callback to preview the find & replace.
 *
 * @param array $find_replace
 *   The database row for the operation.
 * @param FedoraObject $object
 *   The Fedora object to preview the change for.
 *
 * @return array
 *   Rendered diff from the Islandora Pretty Text Diff module.
 */
function islandora_find_replace_preview($find_replace, $object) {
  if (module_exists('islandora_pretty_text_diff')) {
    $dsid = $find_replace['dsid'];
    $datastream = $object[$dsid];
    $content1 = $datastream->content;
    $content2 = islandora_find_replace_string_replace(
      $find_replace['find'],
      $find_replace['replacement'],
      $content1);
    if (strpos($datastream->mimetype, 'xml') !== FALSE) {
      return islandora_pretty_text_diff_content(
        htmlentities($content1, ENT_NOQUOTES),
        htmlentities($content2, ENT_NOQUOTES)
      );
    }
    else {
      return islandora_pretty_text_diff_content($content1, $content2);
    }
  }
  else {
    drupal_set_message(t('Islandora Pretty Text Diff is required'), 'warning');
  }
}