os2web_adlib_api.module

<?php

/**
 * @file
 * Implements API to convert files to PDF or HTML through AdLib.
 */

define('ADLIB_TEST_FID', FALSE);
define('ADLIB_HTML_IMG_FOLDER', 'public://bpimg/');

/**
 * Implements hook_menu().
 */
function os2web_adlib_api_menu() {
  // Added fan for requeueing Adlib jobs.
  $items['admin/config/os2web/reprocess'] = array(
    'title' => 'AdLib requeue',
    'description' => 'Requeue AdLib jobs.',
    'access arguments' => array('administer os2web'),
    'page callback' => 'drupal_get_form',
    'page arguments' => array('os2web_adlib_api_reprocess_form'),
    'type' => MENU_LOCAL_TASK,
    'access callback' => TRUE,
  );
  $items['admin/config/os2web/adlibstatus'] = array(
    'title' => 'AdLib Status',
    'description' => 'Provides an overview of known AdLib jobs.',
    'access arguments' => array('administer os2web'),
    'page callback' => 'os2web_adlib_api_status',
    'type' => MENU_LOCAL_TASK,
  );
  $items['os2web/adlib/ajax/status/%'] = array(
    'page callback' => 'os2web_adlib_api_status_ajax',
    'page arguments' => array(4),
    'type' => MENU_CALLBACK,
    'access arguments' => array('administer os2web'),
  );
  $items['os2web/adlib/test'] = array(
    'title' => 'AdLib test page',
    'access arguments' => array('administer os2web'),
    'page callback' => 'os2web_adlib_api_page',
    'type' => MENU_CALLBACK,
  );
  return $items;
}

/**
 * Implements hook_FORM_alter().
 */
function os2web_adlib_api_form_os2web_settings_settings_form_alter(&$form, &$form_state) {
  // Adlib API configuration.

  $form['os2web_adlib_api_config_group'] = array(
    '#type' => 'fieldset',
    '#title' => t('AdLib Configuration'),
  );
  $form['os2web_adlib_api_config_group']['os2web_adlib_api_allowed_ip_host'] = array(
    '#type' => 'checkbox',
    '#title' => 'Enable AdLib conversion from this site!',
    '#default_value' => variable_get('os2web_adlib_api_is_allowed', 0),
  );
}

/**
 * Ajax callback to update job status.
 *
 * @param int $fid
 *   File identifier.
 */
function os2web_adlib_api_status_ajax($fid) {
  $file = db_select('os2web_adlib_api_doc_files', 'df')
      ->fields('df')
      ->condition('fid', $fid)
      ->execute()
      ->fetchObject();
  if (is_object($file)) {
    if (!in_array($file->job_id, array('', 'Complete', 'Requeue'))) {
      $adlib = _os2web_adlib_api_get_instance();
      $status = $adlib->getJobStatus($file->job_id);
      drupal_json_output($status->status);
    }
    else {
      drupal_json_output('Not queued.');
    }
  }
  else {
    drupal_json_output('Not found.');
  }
}

/**
 * Implements hook_form().
 *
 * Provides form for user to start reprocessing.
 */
function os2web_adlib_api_reprocess_form() {
  $form['submit'] = array(
    '#type' => 'submit',
    '#value' => 'Re-konverter importerede dokumenter',
  );

  return $form;
}

/**
 * Implements hook_form_submit().
 *
 * Starts reprocessing on submit
 */
function os2web_adlib_api_reprocess_form_submit() {
  os2web_adlib_api_reprocess();
}

/**
 * Sets up and starts batchjob.
 */
function os2web_adlib_api_reprocess() {
  $ids = _os2web_adlib_api_bulletpoint_nid();
  foreach ($ids as $nodeid) {
    $operations[] = array('_os2web_adlib_api_reprocess_nid', array($nodeid));
  }
  $batch = array(
    'operations' => $operations,
    'title' => 'Re-konverter adlib dokumenter',
    'init_message' => t('initializing'),
    'progress_message' => t('finished @current of @total'),
    'error_message' => t('something went wrong'),
    'finished' => '_os2web_adlib_api_bulletpoint_finished',
    'file' => drupal_get_path('module', 'os2web_adlib_api') . '/os2web_adlib_api.module',
  );
  batch_set($batch);
  batch_process();
}

/**
 * Provides userfeedback when batchjob is finished.
 * @author Thomas Thune Hansen <tth@bellcom.dk>
 */
function _os2web_adlib_api_bulletpoint_finished($success, $results, $operations) {
  error_log("reschedule finished");
  drupal_set_message(t('All bullet points have been rescheduled for conversion'));
}

/**
 * Fetch all nid's for bulletspoints.
 * @author Thomas Thune Hansen <tth@bellcom.dk>
 *
 * @return array
 *   List of bulletpoint nids
 */
function _os2web_adlib_api_bulletpoint_nid() {
  $query = new EntityFieldQuery();
  $result = $query->entityCondition('entity_type', 'node')
      ->propertyCondition('type', 'bullet_point')
      ->execute();

  $nodeid = array();

  // Entity query returns an object containing an array, containing arrays...
  foreach ($result as $item) {
    foreach ($item as $key => $value) {
      $nodeid[] = $key;
    }
  }
  return $nodeid;
}

/**
 * Worker function, adds fid's attached to nid's to adlib queue.
 * @author Thomas Thune Hansen <tth@bellcom.dk>
 *
 * @param array $nid
 *   List of bulletpoint nids
 * @param array $context
 *   Context information
 */
function _os2web_adlib_api_reprocess_nid($nid, &$context = array()) {
  if (is_array($nid)) {
    foreach ($nid as $n) {
      _os2web_adlib_api_reprocess_nid($n);
    }
    return;
  }
  $node = node_load($nid);

  if (isset($node->field_enclosures['und'])) {
    foreach ($node->field_enclosures['und'] as $file) {
      if (file_exists(drupal_realpath($file['uri'])) && _os2web_adlib_api_is_filetype_approved($file['fid'])) {
        if (os2web_adlib_api_convert_to_pdf($file['fid'])) {
          watchdog('adlib_api', 'readded fid to conversion queue:' . $file['fid'], NULL, WATCHDOG_DEBUG);
        }
      }
    }
  }
}

/**
 * Provides an overview of known AdLib jobs.
 */
function os2web_adlib_api_status() {
  if (isset($_GET['delete'])) {
    if (is_numeric($_GET['delete'])) {
      $query = db_delete('os2web_adlib_api_doc_files');
      $query = $query->condition('fid', $_GET['delete']);
      $query->execute();
    }
    drupal_goto('admin/config/os2web/adlibstatus');
  }
  if (isset($_GET['reset'])) {
    $file = db_select('os2web_adlib_api_doc_files', 'df')
      ->fields('df')
      ->condition('fid', $_GET['reset'])
      ->execute()
      ->fetchObject();

    $query = db_delete('os2web_adlib_api_doc_files');
    $query = $query->condition('fid', $_GET['reset']);
    $query->execute();

    if (is_object($file) && !empty($file->did)) {
      os2web_adlib_api_convert_to_pdf($_GET['reset'], $file->did);
    }
    else {
      os2web_adlib_api_convert_to_pdf($_GET['reset']);
    }
    drupal_goto('admin/config/os2web/adlibstatus');
  }
  if (isset($_GET['process'])) {
    // _os2web_adlib_api_do_processing();
    // drupal_goto('admin/config/os2web/adlibstatus');
    // Add set up queue for adlib processing.
    $batch = array(
      'operations' => array(),
      'title' => 'Konvertere dokumenter via. AdLib',
      'init_message' => t('Initializing'),
      'progress_message' => t('Finished @current of @total'),
      'error_message' => t('Something went wrong'),
    );
    $items = db_select('os2web_adlib_api_doc_files', 'f')
        ->fields('f')
        ->condition(db_or()->condition('job_id', 'Complete', '!=')->isNull('job_id'))
        ->execute()
        ->fetchAll();
    $chunk_size = max(min((int) sqrt(count($items)), 5), 1);
    $items = array_chunk($items, $chunk_size);
    foreach ($items as $item) {
      $batch['operations'][] = array('_os2web_adlib_api_process_fid', array($item));
    }
    batch_set($batch);
    batch_process();
  }
  $adlib = _os2web_adlib_api_get_instance();
  $head = array('Connector Name', 'status', 'Jobs');
  $statuslist = $adlib->getConnectorStatusList();
  $rows = array();
  if (is_array($statuslist)) {
    foreach ($statuslist as $status) {
      $rows[] = array(
        $status->ServerDetail['ServerName'],
        ($status->ServerDetail['Enabled'] == 'True' ? 'E' : 'Not e') . 'nabled - ' . $status->ServerInfo['Status'],
        $status->ServerInfo['JobCount'],
      );
    }
  }
  drupal_add_js(drupal_get_path('module', 'os2web_adlib_api') . '/js/os2web_adlib_api.js', 'file');
  $html = '<h2>Connector information</h2>';
  $html .= theme('table', array('header' => $head, 'rows' => $rows));
  $head = array('Fid', 'Job-id', 'Status', 'action');
  $rows = array();
  $count = db_query("SELECT COUNT(*) as count FROM {os2web_adlib_api_doc_files} WHERE job_id IS NULL OR job_id NOT LIKE 'Complete'")->fetch();
  $files = db_query("SELECT * FROM {os2web_adlib_api_doc_files} WHERE job_id IS NULL OR job_id NOT LIKE 'Complete' LIMIT 50");
  foreach ($files as $file) {
    $fdata = file_load($file->fid);
    $rows[] = array(
      l($fdata->uri, file_create_url($fdata->uri)),
      $file->job_id,
      array('data' => 'Looking up..', 'class' => 'status', 'id' => $file->fid),
      l(t('delete'), 'admin/config/os2web/adlibstatus', array('query' => array('delete' => $file->fid))) . ' ' . l(t('reset'), 'admin/config/os2web/adlibstatus', array('query' => array('reset' => $file->fid))),
    );
  }
  $html .= '<h2>Active jobs (first 50)</h2>';
  $html .= 'jobs total: ' . $count->count;
  $html .= theme('table', array('header' => $head, 'rows' => $rows));
  // $html .= l('Delete all', 'admin/config/os2web/adlibstatus', array('query' => array('delete' => 'all')));
  $html .= ' - ' . l(t('Run processing'), 'admin/config/os2web/adlibstatus', array('query' => array('process' => TRUE)));
  return $html;
}

/**
 * Test page for abitrary test usage.
 */
function os2web_adlib_api_page() {
  error_log('MARK - ' . basename(__FILE__) . ':' . __LINE__ . ' in ' . __FUNCTION__ . '()');
  $fid = 236605;
  $nid = 840980;
  // os2web_adlib_api_file_delete($fid);
  // db_delete('os2web_adlib_api_doc_files')->condition('fid', $fid)->execute();
  // os2web_adlib_api_convert_to_pdf($fid);
  // os2web_adlib_api_convert_to_html($fid, $nid);
  // _os2web_adlib_api_process_pdf();
  // sleep(5);
  // _os2web_adlib_api_process_pdf();
  // $file = db_select('os2web_adlib_api_doc_files', 'df')->fields('df')->condition('fid', $fid)->execute()->fetchObject();
  // $adlib = _os2web_adlib_api_get_instance();
  // $call = array();
  // $call[] = explode('\\',$adlib->addJobFolder());
  // $call[] = $adlib->deleteJobFolder($call[0][1]);
  // $call = $adlib->getJobStatus($file->job_id);
  // $call= file_load(ADLIB_TEST_FID);
  // $call = node_load($nid,NULL,TRUE);
  $files = db_select('file_managed', 'fm')->fields('fm')->execute();
  $rows = array();
  foreach ($files as $file) {
    if (file_exists(drupal_realpath($file->uri))) {
      $rows[] = array(l($file->fid, 'admin/config/os2web/adlibstatus', array(
        'query' => array('reset' => $file->fid))),
        $file->filename);
    }
  }
  return theme('table', array('rows' => $rows));
}

/**
 * Implements hook_cron().
 */
function os2web_adlib_api_cron() {
  // Add set up queue for adlib processing.
  _os2web_adlib_api_do_processing();

  // Add set up queue for adlib requeueing.
  $queue = DrupalQueue::get('os2web_adlib_reprocess');
  if ($queue->numberOfItems() == 0) {
    $items = _os2web_adlib_api_bulletpoint_nid();
    $chunk_size = max(min((int) sqrt(count($items)), 25), 1);
    $items = array_chunk($items, $chunk_size);
    foreach ($items as $item) {
      $queue->createItem($item);
    }
  }
}

/**
 * Implements hook_cron_queue_info().
 * @author Thomas Thune Hansen <tth@bellcom.dk>
 */
function os2web_adlib_api_cron_queue_info() {
  $queues['os2web_adlib_reprocess'] = array(
    'worker callback' => '_os2web_adlib_api_reprocess_nid',
    'time' => 300,
  );
  $queues['os2web_adlib_process'] = array(
    'worker callback' => '_os2web_adlib_api_process_fid',
    'time' => 300,
  );
  return $queues;
}

/**
 * Implements hook_init().
 */
function os2web_adlib_api_init() {
  // _os2web_adlib_api_process_pdf(1);
}

/**
 * Converts a fid to html and injects into nid body field.
 *
 * @param int $fid
 *   File id
 * @param int $nid
 *   Node id
 */
function os2web_adlib_api_convert_to_html($fid, $nid) {
  if (0 === db_query('SELECT fid from {os2web_adlib_api_doc_files} where fid = :fid AND jobtype = :jobtype', array(':fid' => $fid, ':jobtype' => 1))->rowCount()) {
    $record = array(
      'fid' => $fid,
      'did' => $nid,
      'jobtype' => 1,
    );
    drupal_write_record('os2web_adlib_api_doc_files', $record);
    return TRUE;
  }
  else {
    // error_log(basename(__FILE__) . ':' . __LINE__ . " jm@bellcom.dk: File already in queue, aborting.");
    return FALSE;
  }
}

/**
 * Main call to queue file for converting.
 *
 * @param int $fid
 *   File id of a managed file
 * @param int $docref
 *   Arbitrary identifier for later reference (ie. Doc id)
 *
 * @return boolean
 *   False if file was not queued.
 */
function os2web_adlib_api_convert_to_pdf($fid, $docref = NULL) {
  $file = file_load($fid);
  if (!is_object($file)) {
    // File has been erased from drupals file_managed table.
    watchdog('adlib_api', 'File fid: %fid no longer exists, so removed.', array('%fid' => $fid), WATCHDOG_WARNING);
    db_delete('os2web_adlib_api_doc_files')->condition('fid', $fid)->execute();
    return FALSE;
  }
  if (!file_exists(drupal_realpath($file->uri))) {
    watchdog('adlib_api', 'Cannot queue file %file (fid: %fid). File does not exist. File record removed.', array('%fid' => $fid, '%file' => $file->filename), WATCHDOG_WARNING);
    // file_delete($file);
    return FALSE;
  }
  if (_os2web_adlib_api_is_filetype_approved($fid)) {
    if (0 === db_query('SELECT fid from {os2web_adlib_api_doc_files} where fid = :fid AND jobtype = :jobtype', array(':fid' => $fid, ':jobtype' => 0))->rowCount()) {
      $record = array(
        'fid' => $fid,
        'did' => $docref,
        'jobtype' => 0,
      );
      db_delete('os2web_adlib_api_doc_files')
        ->condition('fid', $fid)
        ->condition('did', $docref)
        ->condition('jobtype', 0)
        ->execute();
      drupal_write_record('os2web_adlib_api_doc_files', $record);
      if (ADLIB_TEST_FID) {
        error_log(basename(__FILE__) . ':' . __LINE__ . " jm@bellcom.dk: File queued for processing.");
      }
      return TRUE;
    }
    else {
      // error_log(basename(__FILE__) . ':' . __LINE__ . " jm@bellcom.dk: File already in queue, aborting.");
      return FALSE;
    }
  }
  else {
    // error_log(basename(__FILE__) . ':' . __LINE__ . " jm@bellcom.dk: File already PDF.");
    $record = array(
      'fid' => $fid,
      'did' => $docref,
      'job_id' => 'Complete',
    );
    db_delete('os2web_adlib_api_doc_files')
      ->condition('fid', $fid)
      ->condition('did', $docref)
      ->condition('jobtype', 0)
      ->execute();
    drupal_write_record('os2web_adlib_api_doc_files', $record);
    return FALSE;
  }
}

/**
 * Return the fid of a file saved with a reference.
 *
 * @param int $ref
 *   Reference for the file
 *
 * @return int
 *   File id, false if none.
 */
function os2web_adlib_api_get_fid($ref) {

  $fid = db_query('SELECT fid FROM {os2web_adlib_api_doc_files} WHERE did=:did ORDER BY fid DESC', array(':did' => $ref))->fetchField();
  if ($fid) {
    return $fid;
  }
  return false;
}

/**
 * Factory for class instances. This allows easy changing between api classes.
 *
 * @return AdLibWS
 *   AdLib controller object.
 */
function _os2web_adlib_api_get_instance() {
  $obj = &drupal_static(__FUNCTION__);
  if (is_object($obj)) {
    return $obj;
  }
  module_load_include('inc', 'os2web_adlib_api', 'AdLibWS.class');
  return $obj = new AdLibWS();
}

/**
 * This will run the list of files and call the neccesary workers.
 */
function _os2web_adlib_api_do_processing() {
  // Add set up queue for adlib processing.
  //
  // Disable auto queue of missing files
  $queue = DrupalQueue::get('os2web_adlib_process');
  if ($queue->numberOfItems() == 0) {
    $items = db_select('os2web_adlib_api_doc_files', 'f')
        ->fields('f')
        ->condition(db_or()->condition('job_id', 'Complete', '!=')->isNull('job_id'))
        ->execute()
        ->fetchAll();
    $chunk_size = max(min((int) sqrt(count($items)), 25), 1);
    $items = array_chunk($items, $chunk_size);
    foreach ($items as $item) {
      if ($item->job_id != 'Complete') {
        $queue->createItem($item);
      }
    }
  }
}

/**
 * Worker process for a single file structure.
 *
 * @access: protected
 */
function _os2web_adlib_api_process_fid($file) {
  if (is_array($file)) {
    foreach ($file as $f) {
      _os2web_adlib_api_process_fid($f);
    }
    return;
  }
  if (!is_object($file)) {
    error_log("Var: \$file = " . print_r($file, 1));
    error_log('MARK - ' . basename(__FILE__) . ':' . __LINE__ . ' - Wrong input for fid processing.');
    return;
  }
  if (!lock_acquire(__FUNCTION__ . $file->fid)) {
    lock_wait(__FUNCTION__ . $file->fid, 15);
    return _os2web_adlib_api_process_fid($file);
  }
  if ($file->job_id == NULL || $file->job_id == 'Requeue') {
    _os2web_adlib_api_queue_file($file);
  }
  elseif ($file->job_id != 'Complete') {
    _os2web_adlib_api_process_file($file);
  }
  lock_release(__FUNCTION__ . $file->fid);
}

/**
 * Worker process for an actual file.
 *
 * @access: protected
 */
function _os2web_adlib_api_process_file($file) {
  if (in_array($file->job_id, array('Complete', 'Requeue'))) {
    // We should not get here, but if we do, these posts would be deleted if we do not return here.
    return;
  }
  $adlib = _os2web_adlib_api_get_instance();
  $jobinfo = $adlib->getJobStatus($file->job_id);
  switch ($jobinfo->status) {
    case 'Not found':
    case 'Failed':
      os2web_adlib_api_destroy_job($file->job_id);

      $jobfileinfo = unserialize($file->job_info);

      $fdata = file_load($file->fid);
      if (!file_exists(drupal_realpath($fdata->uri)) || file_uri_scheme($fdata->uri) == '' || $fdata->filemime == 'application/pdf') {
        if ($fdata->filemime == 'application/pdf') {

          $record = array(
            'fid' => $fdata->fid,
            'jobtype' => $fdata->jobtype,
            'job_id' => 'Complete',
            'job_info' => NULL,
          );
          drupal_write_record('os2web_adlib_api_doc_files', $record, array('fid', 'jobtype'));
          watchdog('adlib_api', 'File %file (fid: %fid ) already pdf.', array('%fid' => $fdata->fid, '%file' => $fdata->filename), WATCHDOG_WARNING);
        }
        else {
          watchdog('adlib_api', 'Cannot re-queue file %file. File does not exist or is invalid.', array('%file' => $fdata->filename), WATCHDOG_WARNING);
          os2web_adlib_api_file_delete($file);
        }
      }
      else {
        $record = array(
          'fid' => $file->fid,
          'jobtype' => $file->jobtype,
          'job_id' => 'Requeue',
          'job_info' => NULL,
        );
        drupal_write_record('os2web_adlib_api_doc_files', $record, array('fid', 'jobtype'));
        watchdog('adlib_api', 'Found a failed job for %file. Resetting.', array('%file' => $fdata->filename), WATCHDOG_WARNING);
      }
      break;

    case 'Complete':
      $jobfileinfo = unserialize($file->job_info);

      $download = curl_init($jobfileinfo['out_url']);
      if (variable_get('os2web_pws_proxy')) {
        curl_setopt($download, CURLOPT_PROXY, variable_get('os2web_pws_proxy_host'));
        curl_setopt($download, CURLOPT_PROXYPORT, variable_get('os2web_pws_proxy_port'));
        curl_setopt($download, CURLOPT_HTTPPROXYTUNNEL, TRUE);
      }
      curl_setopt($download, CURLOPT_RETURNTRANSFER, TRUE);
      $data = curl_exec($download);

      $info = curl_getinfo($download);
      if ($info['http_code'] !== 200) {
        // File not found ??
        $record = array(
          'fid' => $file->fid,
          'jobtype' => $file->jobtype,
          'job_id' => 'Requeue',
          'job_info' => NULL,
        );
        os2web_adlib_api_destroy_job($file->job_id);
        drupal_write_record('os2web_adlib_api_doc_files', $record, array('fid', 'jobtype'));
        watchdog('adlib_api', 'Completed file (fid: %fid)could not be found on the server (Code: %code). Job reset.', array('%code' => $info['http_code'], '%fid' => $file->fid), WATCHDOG_WARNING);
        break;

      }

      switch ($file->jobtype) {
        // PDF.
        case 0:
          $upd_file = file_load($file->fid);
          $uri = file_unmanaged_save_data($data, str_replace($upd_file->filename, $jobfileinfo['out_filename'], $upd_file->uri), FILE_EXISTS_REPLACE);
          if (filesize(drupal_realpath($uri)) > 0) {
            db_delete('file_managed')->condition('uri', $uri)->execute();
            $upd_file->uri = $uri;
            $upd_file->filename = basename($uri);
            $upd_file->filemime = file_get_mimetype($upd_file->uri);
            $upd_file->filesize = filesize(drupal_realpath($upd_file->uri));
            $upd_file->timestamp = time();
            os2web_adlib_api_destroy_job($file->job_id);
            $record = array(
              'fid' => $file->fid,
              'jobtype' => $file->jobtype,
              'job_id' => 'Complete',
            );
            drupal_write_record('os2web_adlib_api_doc_files', $record, array('fid', 'jobtype'));
            drupal_write_record('file_managed', $upd_file, array('fid'));
            field_cache_clear();
            watchdog('adlib_api', 'File (fid: %fid) was converted successfully to pdf.', array('%fid' => $file->fid), WATCHDOG_INFO);
            error_log(basename(__FILE__) . ':' . __LINE__ . " jm@bellcom.dk: File saved, file record updated. New URI: " . $uri);
          }
          else {
            error_log(basename(__FILE__) . ':' . __LINE__ . " jm@bellcom.dk: Filesize was 0. Aborting update.");
          }
          break;

        // HTML.
        case 1:
          $nid = $file->did;
          if ($node = node_load($nid)) {
            $node->field_bpa_body[LANGUAGE_NONE][0]['value'] = _os2web_adlib_api_process_html($data, $jobfileinfo['out_url']);
            $node->field_bpa_body[LANGUAGE_NONE][0]['format'] = 'filtered_html';
            $node->feeds = array();
            node_save($node);
            os2web_adlib_api_destroy_job($file->job_id);
            $record = array(
              'fid' => $file->fid,
              'jobtype' => $file->jobtype,
              'job_id' => 'Complete',
            );
            drupal_write_record('os2web_adlib_api_doc_files', $record, array('fid', 'jobtype'));
            error_log('MARK - ' . basename(__FILE__) . ':' . __LINE__ . ' in ' . __FUNCTION__ . '() - File converted, node updated.');
            watchdog('adlib_api', 'File (fid: %fid) was converted successfully.', array('%fid' => $file->fid), WATCHDOG_INFO);
          }
          break;

        default:
          watchdog('adlib_api', 'Unknown jobtype returned from AdLib.', NULL, WATCHDOG_WARNING);
      }
      break;

    case 'Pending':
    case 'InProgress':
      break;

    default:
      watchdog('adlib_api', 'Unknown job status: %var', array('%var' => $jobinfo->status), WATCHDOG_WARNING);
      break;

  }
}

/**
 * This function validates html and fixes all image links.
 * @access: protected
 */
function _os2web_adlib_api_process_html($data, $base_url) {
  $base_dir = ADLIB_HTML_IMG_FOLDER;
  if (!file_prepare_directory($base_dir, FILE_CREATE_DIRECTORY | FILE_MODIFY_PERMISSIONS)) {
    watchdog('adlib_api', 'Cannot get write permission for the AdLib image store folder.', NULL, WATCHDOG_ERROR);
    return;
  }
  module_load_include('php', 'os2web_adlib_api', 'simple_html_dom');
  $html = str_get_html($data);
  $img = $html->find('img');
  for ($i = 0; $i < count($img); $i++) {
    watchdog('adlib_api', 'Importing image for HTML: %fname', array('%fname' => basename($img[$i]->src)));
    $image_url = $base_url . '/' . rawurlencode($img[$i]->src);
    $download = curl_init($image_url);
    if (variable_get('os2web_pws_proxy')) {
      curl_setopt($download, CURLOPT_PROXY, variable_get('os2web_pws_proxy_host'));
      curl_setopt($download, CURLOPT_PROXYPORT, variable_get('os2web_pws_proxy_port'));
      curl_setopt($download, CURLOPT_HTTPPROXYTUNNEL, TRUE);
    }
    curl_setopt($download, CURLOPT_RETURNTRANSFER, TRUE);
    $image = curl_exec($download);

    $info = curl_getinfo($download);
    if ($info['http_code'] === 200) {
      // AdLib uses Windows folder notation.
      $file_uri = $base_dir . basename(str_replace('\\', '/', $img[$i]->src));
      $file = file_save_data($image, $file_uri, FILE_EXISTS_REPLACE);
      $img[$i]->src = file_create_url($file->uri);
    }
    else {
      watchdog('adlib_api', 'Failed to fetch image for HTML: %fname', array('%fname' => basename($element->src)), WATCHDOG_WARNING);
    }
  }
  $data = (string) $html;

  // Clean up the html, since AdLib sometimes return less-than-valid html and hence can cause PDO exceptions.
  if (!function_exists('tidy_repair_string')) {
    watchdog('adlib_api', 'Tidy API not found! Cannot process HTML.', NULL, WATCHDOG_ERROR);
  }
  else {
    $config = array(
      'indent' => TRUE,
      'output-xhtml' => TRUE,
      'wrap' => 200,
      'word-2000' => TRUE,
      'show-body-only' => TRUE,
    );
    $data = tidy_repair_string($data, $config);
  }
  return $data;
}

/**
 * Worker function that queues up a single file from a row in the job table.
 *
 * @param object $file
 *   Object representing a row from the job table
 *
 * @return boolean
 *   True on success
 */
function _os2web_adlib_api_queue_file($file) {
  $file->file = file_load($file->fid);

  if (is_file(drupal_realpath($file->file->uri)) && _os2web_adlib_api_is_filetype_approved($file->file->fid)) {
    $adlib = _os2web_adlib_api_get_instance();
    $upload = $adlib->uploadData($file->file->filename, file_get_contents(drupal_realpath($file->file->uri)));
    if ($upload) {
      // Normal priority.
      $priority = 3;
      if ($file->job_id === 'Requeue') {
        // Make requeued jobs run on lower priority.
        $priority = 5;
      }
      $job = $adlib->addJob($upload, $file->jobtype, $priority);
      if ($job !== FALSE) {
        $jobinfo = array(
          'fid' => $file->fid,
          'jobtype' => $file->jobtype,
          'job_id' => (string) $job->JobSettings['JobID'],
          'job_info' => $upload,
        );
        drupal_write_record('os2web_adlib_api_doc_files', $jobinfo, array('fid', 'jobtype'));
        watchdog('adlib_api', 'Queued file %file for PDF.', array('%file' => $file->file->filename));
        return TRUE;
      }
      else {
        watchdog('adlib_api', 'Failed to add job for  %file.', array('%file' => $file->file->filename));
      }
    }
    else {
      watchdog('adlib_api', 'Upload of %file failed. Check permissions.', array('%file' => $file->file->filename));
      db_delete('os2web_adlib_api_doc_files')->condition('fid', $file->file->fid)->execute();
    }
  }
  else {
    if (!_os2web_adlib_api_is_filetype_approved($file->file->fid)) {
      $jobinfo = array(
        'fid' => $file->fid,
        'jobtype' => $file->jobtype,
        'job_id' => 'Complete',
        'job_info' => NULL,
      );
      drupal_write_record('os2web_adlib_api_doc_files', $jobinfo, array('fid', 'jobtype'));
      watchdog('adlib_api', 'Failed to add %file to adlib queue. File is already pdf.', array('%file' => $file->file->filename));
    }
    else {
      watchdog('adlib_api', 'Failed to add %file to adlib queue. File not found.', array('%file' => $file->file->filename));
      db_delete('os2web_adlib_api_doc_files')->condition('fid', $file->file->fid)->execute();
    }
  }
  return FALSE;
}

/**
 * Helper function, check if file type is correct.
 * @access protected
 *
 * @param int $fid
 *   File id
 *
 * @return bool
 *   True if file is approved for processing.
 */
function _os2web_adlib_api_is_filetype_approved($fid) {

  // Filetypes which are approved for processing.
  $approved_files = array(
    'docx',
    'docm',
    'doc',
    'dotx',
    'dotm',
    'xps',
    'rtf',
    'xlsx',
    'odt',
    'xlsm',
    'xlsb',
    'xls',
    'xltx',
    'xltm',
    'xlt',
    'xps',
    'ods',
    'wps',
    'pptx',
    'pptm',
    'ppt',
    'potx',
    'potm',
    'ppsx',
    'ppsm',
    'pps',
    'odp',
    'msg',
  );

  if (is_array($fid)) {
    $result = array();
    $files = file_load_multiple($fid);
    foreach ($files as $file) {
      $result[$file->fid]['is_pdf'] = in_array(strtolower(pathinfo(file_load($fid)->filename, PATHINFO_EXTENSION)), $approved_files);
      $result[$file->fid]['file'] = $file;
    }
    return $result;
  }
  else {
    return in_array(strtolower(pathinfo(file_load($fid)->filename, PATHINFO_EXTENSION)), $approved_files);
  }
}

/**
 * Deletes a job related to a file.
 *
 * @param object $file
 *   Drupal file object
 */
function os2web_adlib_api_file_delete($file) {
  if (is_numeric($file)) {
    $file = db_select('os2web_adlib_api_doc_files', 'f')
        ->fields('f')
        ->condition('fid', $file)
        ->execute()
        ->fetchObject();
  }
  if (is_object($file)) {
    if (isset($file->job_id)) {
      os2web_adlib_api_destroy_job($file->job_id);
    }
    db_delete('os2web_adlib_api_doc_files')->condition('fid', $file->fid)->execute();
    watchdog('adlib_api', 'File %file has been cleaned up from adlib conversion table.', array('%file' => $file->uri), WATCHDOG_DEBUG);
  }
}

/**
 * Removes the job and related folders from the AdLib server.
 *
 * @param string $job_id
 *   Job id
 */
function os2web_adlib_api_destroy_job($job_id) {
  $file = db_select('os2web_adlib_api_doc_files', 'f')
      ->fields('f')
      ->condition('job_id', $job_id)
      ->execute()
      ->fetchObject();
  if (is_object($file)) {
    $adlib = _os2web_adlib_api_get_instance();
    if (isset($file->job_id) && $file->job_id != 'Complete') {
      $jobfileinfo = unserialize($file->job_info);
      $adlib->deleteJob($file->job_id);
      $folder = explode('\\', $jobfileinfo['basefolder']);
      $adlib->deleteJobFolder($folder[1]);
    }
    watchdog('adlib_api', 'Deleted job for %fid', array('%fid' => $job_id));
  }
}

/**
 * Implements hook_os2web_help().
 */
function os2web_adlib_api_os2web_help($sections) {

  // List of content.
  $sections['list_of_content'] = t('<a href="#os2web_adlib_api">Adlib file conversion</a><br />');

  // Module specific.
  $sections['os2web_adlib_api'] = t('<h2 id="os2web_adlib_api">Adlib file conversion:</h2>');
  $sections['os2web_adlib_api'] .= t('<p><b>Description:</b><br />Adlib is a file conversion unit on your network. This module is often used to convert misc files to either .html or .pdf files.<br />See the status of the current queue of files at the <a href="@status" target="_blank">adlib status page</a>.<br />You can requeue files for conversion at the <a href="@requeue" target="_blank">requeue page</a>.</p>', array('@status' => url('admin/config/os2web/adlibstatus'), '@requeue' => url('admin/config/os2web/reprocess')));
  $sections['os2web_adlib_api'] .= t('<p><b>Configuration:</b><br /> Setup your AdLib server at the <a href="@settings" target="_blank">OS2web Settings Page</a>. Enter the AdLib service endpoint address.</p>', array('@settings' => url('admin/config/os2web/adlibstatus')));

  return $sections;
}