feedapi_imagegrabber.module

<?php

//(c) Srijan Technologies Pvt. Ltd. 2009

/**
 * @file
 * Grabs the primary image from the feed-item and
 * stores into a CCK imagefield.
 */

// Number of feeds to process for each step in cron.
define('FEEDAPI_IMAGEGRABBER_CRON_FEEDS', 20);

/**
 * Implementation of hook_help().
 */
function feedapi_imagegrabber_help($path, $arg) {
  switch ($path) {
    case 'admin/help#feedapi_imagegrabber':
      $output = '<p>'. t('Provides a parser for <a href="@feedapi">FeedAPI</a>, which visits the original posts of a feed and downloads the main image from the post.', array('@feedapi' => url('admin/help/feedapi'))) .'</p>';
      $output .= '<p>'. t('FeedAPI ImageGrabber mimics the thumbnail display of Google Reader on your Drupal website.') .'</p>';
      $output .= '<p>'. t('After downloading the main image from the post, it is stored with the node created by FeedAPI Node, inside an Imagefield. You can add new imagefields to a content type by clicking <i>manage fields</i> on the <a href="@content-types">content types</a> page. You can also use ImageCache and ImageAPI to add presets to the imagefields', array('@content-types' => url('admin/content/types'))) .'</p>';
      $output .= '<p>'. t('Usually ImageGrabber will search the whole of the web-page for an appropriate image, but you can restrict its search by specifying the id/class of the tag between which, you think, is the appropriate image. Please look <a href="@id-class">here</a> for a tutorial on id/class', array('@id-class' => url('http://www.w3.org/TR/html401/struct/global.html#h-7.5.2'))) .'</p>';
      $output .= '<p>'. t('I have prepared a turorial for FeedAPI ImageGrabber. Please look <a href="@tutorial">here</a> for help. Please open issues in the support forums if you face any issues.', array('@tutorial' => url('http://publicmind.in/blog/tutorial-for-feedapi-imagegrabber'))) .'</p>';
      return $output;
  }
}

/**
 * Implementation of hook_menu().
 */
function feedapi_imagegrabber_menu() {
  $items = array();
  $items['node/%node/grabber'] = array(
    'title' => 'Grab Images',
    'page callback' => 'feedapi_imagegrabber_download_images',
    'page arguments' => array(1),
    'type' => MENU_LOCAL_TASK,
    'access callback' => '_feedapi_imagegrabber_op_access',
    'access arguments' => array(1),
  );

  return $items;
}

/**
 * helper access function for hook_menu().
 */
function _feedapi_imagegrabber_op_access($node) {
  $enabled = feedapi_imagegrabber_is_enabled($node->nid);
  if (!isset($enabled) || $enabled == 0 || $enabled == '0') {
    return FALSE;
  }
  global $user;
  $own_feed = $node->uid == $user->uid && user_access('edit own '. $node->type .' content') ? TRUE : FALSE;
  return user_access('administer imagegrabber') || $own_feed;
}

/**
 * Implementation of hook_perm().
 */
function feedapi_imagegrabber_perm() {
  return array('administer imagegrabber', 'enable imagegrabber');
}

/**
 * validates and uploads the image file into the cck image field
 *
 * @param   $filename   path to the image to be uploaded
 * @param   $field      cck image field instance to insert into which image is
 *                      uploaded (check before passing that this field exists)
 *
 * @return
 * fid of the image if the image is uploaded successfuly,
 * FALSE if any error occurs.
 */
function feedapi_imagegrabber_image_upload($filename, $field, $replace = FILE_EXISTS_RENAME) {

  if (empty($filename)) {
    return 0;
  }

  $field_instance = content_fields($field);
  $dest = filefield_widget_file_path($field_instance);

  if (!field_file_check_directory($dest, FILE_CREATE_DIRECTORY)) {
    drupal_set_message(t('ImageGrabber: %dir directory is not writable, Please check the permissions', array('%dir' => $dest)), 'error', FALSE);
    return 0;
  }

  global $user;

  $validators1 = imagefield_widget_upload_validators($field_instance);
  $validators2 = filefield_widget_upload_validators($field_instance);
  $validators  = array_merge($validators1, $validators2);
  // Add in our check of the the file name length.
  $validators['file_validate_name_length'] = array();
  $image_extensions = 'jpg jpeg gif png';

  // Begin building file object.
  $file           = new stdClass();
  $file->filename = file_munge_filename(trim(basename($filename)), $image_extensions);
  $file->filepath = $filename;
  $file->filemime = file_get_mimetype($file->filename);


  // Rename potentially executable files, to help prevent exploits.
  if (preg_match('/\.(php|pl|py|cgi|asp|js)$/i', $file->filename) && (substr($file->filename, -4) != '.txt')) {
    return 0;
  }

  $file->source = $field ."_imagegrabber";

  $destination_path = file_destination(file_create_path($dest .'/'. $file->filename), $replace);
  $path_original = $filename;
  if (file_copy($filename, $destination_path, FILE_EXISTS_REPLACE)) {
    if (!($path_original == $destination_path || file_delete($path_original))) {
      drupal_set_message(t('ImageGrabber: Please remove the file %file, ImageGrabber was unable to do so.', array('%file' => $path_original)), 'warning');
    }
  }
  else {
    return 0;
  }

  $filename          = $destination_path;
  $file->filename    = file_munge_filename(trim(basename($filename)), $image_extensions);
  $file->destination = $filename;
  $file->filepath    = $filename;
  chmod($filename, 0744);
  if (!($filesize = filesize($filename))) {
    drupal_set_message(t('ImageGrabber: Filesize for some files couldn\'t be calculated and aren\'t saved.'), 'error', FALSE);
    file_delete($filename);
    return 0;
  }

  $file->filesize = $filesize;

  // Call the validation functions.
  $errors = array();
  foreach ($validators as $function => $args) {
    array_unshift($args, $file);
    $errors = array_merge($errors, call_user_func_array($function, $args));
  }

  // Check for validation errors.
  if (!empty($errors)) {
    $message = t('ImageGrabber: The selected file %name could not be uploaded.', array('%name' => $file->filename));
    if (count($errors) > 1) {
      $message .= '<ul><li>'. implode('</li><li>', $errors) .'</li></ul>';
    }
    else {
      $message .= ' '. array_pop($errors);
    }
    drupal_set_message($message, 'error', FALSE);
    file_delete($filename);
    return 0;
  }

  $file->uid       = $user->uid;
  $file->status    = FILE_STATUS_TEMPORARY;
  $file->timestamp = time();

  drupal_write_record('files', $file);


  // Let modules add additional properties to the yet barebone file object.
  foreach (module_implements('file_insert') as $module) {
    $function = $module .'_file_insert';
    $function($file);
  }

  return $file;
}

/**
 * Implementation of hook_form_alter().
 */
function feedapi_imagegrabber_form_alter(&$form, $form_state, $form_id) {

  if (user_access('enable imagegrabber')) {
    if (isset($form['type']) && isset($form['#node']) && $form['type']['#value'] .'_node_form' == $form_id && feedapi_enabled_type($form['type']['#value'])) {

      // Don't blow away existing form elements.
      if (!isset($form['imagegrabber'])) {
        $form['imagegrabber'] = array();
      }
      $form['imagegrabber'] += array(
        '#type' => 'fieldset',
        '#title' => t('ImageGrabber'),
        '#collapsible' => TRUE,
        '#collapsed' => FALSE,
        '#tree' => TRUE,
      );

      $node = $form['#node'];
      if (isset($node->nid)) {
        $feedapi_settings = feedapi_get_settings($node->type, $node->vid);
        if (isset($feedapi_settings['processors']['feedapi_node'])) {
          $is_feedapi_node_enabled = TRUE;
        }
        else $is_feedapi_node_enabled = FALSE;
      }
      else {
        $feedapi_settings = feedapi_get_settings($node->type);
        if (isset($feedapi_settings['processors']['feedapi_node']['enabled'])) {
          $is_feedapi_node_enabled = $feedapi_settings['processors']['feedapi_node']['enabled'];
        }
        else $is_feedapi_node_enabled = FALSE;
      }

      if (!$is_feedapi_node_enabled) {
        $form['imagegrabber']['value'] = array(
          '#type' => 'value',
          '#value' => $is_feedapi_node_enabled,
        );

        $form['imagegrabber']['message'] = array(
          '#title' => t('Enable the feedapi_node processor for this feed to use Imagegrabber'),
          '#type' => 'item',
          '#description' => t('ImageGrabber settings have been removed, because it needs the feedapi_node processor. Enable the feedapi_node processor for this feed to use ImageGrabber.'),
        );

        return;
      }

      if (!isset($node->nid) || feedapi_imagegrabber_get_settings($node->nid) == FALSE) {

        $enabled = 0;
        $image_field = '0';
        $id_class = 0;
        $id_class_description = '';
        $feeling_lucky = 0;
      }
      else {

        $imagegrabber_settings = feedapi_imagegrabber_get_settings($node->nid);
        $enabled = $imagegrabber_settings['enabled'];
        $image_field = $imagegrabber_settings['image_field'];
        $id_class = $imagegrabber_settings['id_class'];
        $id_class_description = $imagegrabber_settings['id_class_description'];
        $feeling_lucky = $imagegrabber_settings['feeling_lucky'];
      }

      $content_type_processor = $feedapi_settings['processors']['feedapi_node']['content_type'];
      $options = array(t('Select an Image field'),
      );

      if (isset($content_type_processor) && !empty($content_type_processor)) {
        $type = content_types($content_type_processor);
        $fields = $type['fields'];
        foreach ($fields as $type => $args) {
          if ($args['widget']['type'] == 'imagefield_widget') {
            $imagefield = $args['field_name'];
            $options[$imagefield] = $args['field_name'];
          }
        }
        if (count($options) == 1) {
          $form['imagegrabber']['value'] = array(
            '#type' => 'value',
            '#value' => 0,
          );

          $form['imagegrabber']['message'] = array(
            '#title' => t('There are no imagefields associated with <i>@node</i> content type.', array('@node' => $content_type_processor)),
            '#type' => 'item',
            '#description' => t('ImageGrabber stores the image in an image field of the feed item node. The content type for the feed-item of this feed is <i>@node</i>. Add Imagefields to <i>@node</i> content type in order to enable ImageGrabber.', array('@node' => $content_type_processor)),
          );

          return;
        }
      }
      else {
        $form['imagegrabber']['value'] = array(
          '#type' => 'value',
          '#value' => 0,
        );

        $form['imagegrabber']['message'] = array(
          '#title' => t('ImageGraber could not find the associated content type for the feed-item. Please try re-loading the page, otherwise report on the support forums.'),
          '#type' => 'item',
          '#description' => t('This is an un-expected error. It might be due to change in APIs of other dependent modules. Please report the issue at the earliest.'),
        );
        return;
      }

      $form['imagegrabber']['content_type_node'] = array(
        '#type' => 'value',
        '#value' => $content_type_processor,
      );

      $form['imagegrabber']['enabled'] = array(
        '#type' => 'checkbox',
        '#title' => t('Enable ImageGrabber for this feed'),
        '#description' => t('Check if you want to download the images of feed items to your site.'),
        '#default_value' => $enabled,
        '#weight' => -17,
      );

      if ($enabled == 1 || $enabled == '1') {
        $default = $image_field;
      }
      else {
        $default = $options[0];
      }

      $form['imagegrabber']['image_field'] = array(
        '#type' => 'select',
        '#title' => t('Store the image in: (Select an Imagefield)'),
        '#description' => t('ImageGrabber stores the image into this image field of the feed item. The content type for the feed-item of this feed is <i>@node</i>', array('@node' => $content_type_processor)),
        '#options' => $options,
        '#default_value' => $default,
        '#weight' => -16,
      );

      $form['imagegrabber']['id_class'] = array(
        '#type' => 'radios',
        '#title' => t('Search for an image between the tag which is identified by '),
        '#options' => array(t('None, search the whole web-page for the image.'),
          t('an ID'),
          t('a Class'),
        ),
        '#description' => t('Select <i>None</i> if you are not sure what this means. For help, click <a href="@feedapi-imagegrabber">here</a>. Enter the <i>ID</i> or <i>CLASS</i> of the tag in the textfield below. ', array('@feedapi-imagegrabber' => url('http://publicmind.in/blog/tutorial-for-feedapi-imagegrabber'))),
        '#default_value' => $id_class,
        '#weight' => -15,
      );


      $form['imagegrabber']['id_class_description'] = array(
        '#type' => 'textfield',
        '#title' => t('<i>ID</i> or <i>CLASS</i> of the HTML tag (Leave empty if you selected <i>None</i> above.)'),
        '#description' => t('Enter the <i>ID</i> or <i>CLASS</i> of the html element between which you want ImageGrabber to search for the image. In case of <i>CLASS</i>, the first tag matching the specified <i>CLASS</i> will be selected. Note that both <i>ID</i> and <i>CLASS</i> are CASE-SENSITIVE. For help, click <a href="@feedapi-imagegrabber">here</a>.', array('@feedapi-imagegrabber' => url('http://publicmind.in/blog/tutorial-for-feedapi-imagegrabber'))),
        '#default_value' => $id_class_description,
        '#maxlength' => 100,
        '#weight' => -14,
      );


      $form['imagegrabber']['feeling_lucky'] = array(
        '#type' => 'radios',
        '#title' => t('Feeling Lucky, huh?'),
        '#options' => array(t('No, Select the largest image between the tag.'),
          t('Yes, Select the first image between the tag. (Recommended)'),
        ),
        '#default_value' => $feeling_lucky,
        '#weight' => -13,
      );

      //set the value to 1 before returning the form, because everything went ok.
      $form['imagegrabber']['value'] = array(
        '#type' => 'value',
        '#value' => 1,
      );

      $form['#validate'][] = 'feedapi_imagegrabber_form_node_validate';
    }
  }
}

/**
 * validates the form altered for the feedapi enabled content type node.
 */
function feedapi_imagegrabber_form_node_validate($form, &$form_state) {
  $value = $form_state['values']['imagegrabber']['value'];
  if ($value == 1) {
    $enabled = $form['imagegrabber']['enabled']['#value'];
    $image_field = $form['imagegrabber']['image_field']['#value'];

    if ($enabled == 1) {
      if ($image_field == '0') {
        form_set_error("imagegrabber']['image_field", "ImageGrabber is enabled but no image field is selected");
        return;
      }

      $id_class = $form['imagegrabber']['id_class']['#value'];
      $id_class_desc = $form['imagegrabber']['id_class_description']['#value'];

      if ($id_class != 0) {
        if (!isset($id_class_desc) || empty($id_class_desc) || $id_class_desc == '') {
          form_set_error("imagegrabber']['id_class_description", "ImageGrabber: ID or Class field can not be empty.");
          return;
        }
        if (!preg_match('/^[a-zA-Z]+[_a-zA-Z0-9-]*$/', $id_class_desc)) {
          form_set_error("imagegrabber']['id_class_description", "ImageGrabber: ID/CLASS string is not valid. Only alphabets, digits, hyphens and underscores are allowed.");
          return;
        }
      }

      if (!isset($form_state['values']['feedapi']['processors']['feedapi_node'])) {
        form_set_value($form['imagegrabber']['enabled'], 0, $form_state);
        drupal_set_message(t('ImageGrabber has been disabled for this feed as it depends on feedapi_node processor and it was unfortunately disabled.'), 'warning');
        return;
      }
      $content_type_node_new = $form['feedapi']['processors']['feedapi_node']['content_type']['#value'];

      $content_type_node_old = $form['imagegrabber']['content_type_node']['#value'];
      if (strcmp($content_type_node_new, $content_type_node_old) != 0) {
        form_set_value($form['imagegrabber']['enabled'], 0, $form_state);
        drupal_set_message(t('ImageGrabber has been disabled for this feed becuase you changed the content type of the feed items. Please edit the feed again to enable ImageGrabber'), 'warning');
        return;
      }
    }
  }
}

/*
 * Implementation of hook_nodeapi().
 */
function feedapi_imagegrabber_nodeapi(&$node, $op, $teaser, $page) {
  if (feedapi_enabled_type($node->type)) {
    switch ($op) {
      case 'insert':
      case 'update':
        $value = $node->imagegrabber['value'];
        $nid = $node->nid;
        if ($value && isset($nid)) {
          $enabled = $node->imagegrabber['enabled'];
          if ($enabled == 1) {
            $image_field = $node->imagegrabber['image_field'];
            $id_class = $node->imagegrabber['id_class'];
            if ($id_class != 0) {
              $id_class_desc = $node->imagegrabber['id_class_description'];
            }
            else {
              $id_class_desc = '';
            }
            $feeling_lucky = $node->imagegrabber['feeling_lucky'];
            db_query("UPDATE {feedapi_imagegrabber} SET enabled = %d , image_field = '%s' , id_class = %d , id_class_description = '%s' , feeling_lucky = %d WHERE nid = %d", $enabled, $image_field, $id_class, $id_class_desc, $feeling_lucky, $nid);
            if (!db_affected_rows()) {
              @db_query("INSERT INTO {feedapi_imagegrabber} (nid, enabled, image_field, id_class, id_class_description, feeling_lucky) VALUES (%d, %d, '%s', %d, '%s', %d)", $nid, $enabled, $image_field, $id_class, $id_class_desc, $feeling_lucky);
            }
          }
          else {
            db_query("UPDATE {feedapi_imagegrabber} SET enabled = %d WHERE nid = %d", $enabled, $nid);
          }
        }
        break;

      case 'delete':
        $nid = $node->nid;
        @db_query("DELETE FROM {feedapi_imagegrabber} where nid = %d", $nid);
        @db_query("DELETE FROM {feedapi_imagegrabber_cron_nodes} where feed_nid = %d", $nid);
        break;
    }
  }
}

/**
 * Implementation of hook_feedapi_after_refresh().
 *
 * Called after feed is refreshed. It downloads the images
 * and stores it into the cck image field.
 *
 * @param $feed the feed object which contains all the feed items.
 *        $is_cron if it is called by us or feedapi
 */
function feedapi_imagegrabber_feedapi_after_refresh($feed, $is_cron = FALSE) {

  if (!user_access('administer imagegrabber')) {
    return;
  }

  $feed = (array)$feed;
  if (!isset($feed['nid'])) {
    return;
  }
  $feed_nid = $feed['nid'];
  $settings = feedapi_imagegrabber_get_settings($feed_nid);
  if ($settings == FALSE) {
    return;
  }
  if (!function_exists('curl_exec')) {
    drupal_set_message(t('ImageGrabber: The cURL Library for PHP is missing or outdated. Go to <a href="!admin-reports-status">Status Report page</a>', array('!admin-reports-status' => url('admin/reports/status'))), 'error');
    return;
  }
  $path = drupal_get_path('module', 'feedapi_imagegrabber') .'/url_to_absolute.php';
  if (!file_exists($path)) {
    drupal_set_message(t('ImageGrabber: The url conversion script is missing. Go to <a href="!admin-reports-status">Status Report page</a>', array('!admin-reports-status' => url('admin/reports/status'))), 'error');
    return;
  }
  require_once($path);

  $enabled = $settings['enabled'];
  $image_field = $settings['image_field'];
  $id_class = $settings['id_class'];
  $id_class_description = $settings['id_class_description'];
  $feeling_lucky = $settings['feeling_lucky'];

  if (!isset($enabled) || $enabled == 0 || $enabled == '0') {
    return;
  }

  if ($enabled == 1 && (!isset($image_field) || empty($image_field) || $image_field == '0')) {
    return;
  }

  if (($id_class == 1 || $id_class == 2) && (!isset($id_class_description) || empty($id_class_description) || $id_class_description == '')) {
    return;
  }

  $cron_items = array();
  $cron_empty = TRUE;
  foreach ($feed['items'] as $item => $argument) {
    if (isset($argument->nid)) {
      $nid = $argument->nid;
    }
    else {
      $original_url = $argument->options->original_url;
      if (!isset($original_url) || empty($original_url)) {
        continue;
      }
      $nid = db_fetch_object(db_query("SELECT nid FROM {feedapi_node_item} WHERE url = '%s'", $original_url));
      if (!isset($nid) || empty($nid)) {
        continue;
      }
    }

    if (!$cron_empty || !feedapi_imagegrabber_cron_time()) {
      $cron_items[$nid] = $argument->options->original_url;
      $cron_empty = FALSE;
      continue;
    }
    $original_url = $argument->options->original_url;
    if (!($node = node_load($nid))) {
      continue;
    }
    $type = content_types($node->type);
    $fields = $type['fields'];
    if (!(array_key_exists($image_field, $fields))) {
      continue;
    }
    $filename = feedapi_imagegrabber_select_image($original_url, $id_class, $id_class_description, $feeling_lucky);
    if (!$filename) {
      continue;
    }
    if (!($file = feedapi_imagegrabber_image_upload($filename, $image_field))) {
      continue;
    }
    if (isset($file->fid)) {
      $node->$image_field = array(
        array(
          'fid' => $file->fid,
          'filename' => $file->filename,
          'filepath' => $file->filepath,
          'status' => $file->status,
          'filesize' => $file->filesize,
          'list' => 1,
        ),
      );

      //a little hack to make filefield_paths module <= v1.3 process this node
      if (module_exists('filefield_paths')) {
        $node->form_id = $node->type .'_node_form';
      }

      node_save($node);
    }
  }
  if ($is_cron) {
    @db_query("DELETE FROM {feedapi_imagegrabber_cron_nodes} where feed_nid = %d", $feed_nid);
  }
  if (!$cron_empty) {
    $data = db_fetch_array(db_query("SELECT feed_nid, nid FROM {feedapi_imagegrabber_cron_nodes} WHERE feed_nid = %d", $feed_nid));
    if ($data != FALSE) {
      $data_nid = unserialize($data['nid']);
      foreach ($cron_items as $node => $url) {
        $data_nid[$node] = $url;
      }
      @db_query("UPDATE {feedapi_imagegrabber_cron_nodes} SET nid = '%s' WHERE feed_nid = %d", serialize($data_nid), $feed_nid);
    }
    else {
      @db_query("INSERT INTO {feedapi_imagegrabber_cron_nodes} (feed_nid, nid) VALUES (%d, '%s')", $feed_nid, serialize($cron_items));
    }
    drupal_set_message(t('ImageGrabber: PHP execution time limit for system is !limit seconds, due to which images for some feed items couldn\'t be downloaded. Please click on \'Grab Images\' to refresh those feed-items.', array('!limit' => ini_get('max_execution_time'))), 'warning');
  }
  else {
    drupal_set_message(t('ImageGrabber completed successfully!!'));
  }
}

/**
 * Implementation of hook_cron().
 */
function feedapi_imagegrabber_cron() {
  $process = 0;
  while (!$process && feedapi_imagegrabber_cron_time()) {
    $process = FEEDAPI_IMAGEGRABBER_CRON_FEEDS;
    $result = db_query_range("SELECT feed_nid, nid FROM {feedapi_imagegrabber_cron_nodes}", FEEDAPI_CRON_FEEDS);

    while (feedapi_imagegrabber_cron_time() && $data = db_fetch_array($result)) {
      $feed_items = unserialize($data['nid']);
      $feed_nid   = $data['feed_nid'];
      $feed       = feedapi_imagegrabber_create_feed($feed_nid, $feed_items);
      feedapi_imagegrabber_feedapi_after_refresh($feed, TRUE);
      $process--;
    }
  }
}

/**
 * Check for time limits in feed processing.
 *
 * @return boolean FALSE (time exceeded)
 */
function feedapi_imagegrabber_cron_time() {

  $max_exec_time = ini_get('max_execution_time') == 0 ? 120 : ini_get('max_execution_time') * 2 / 3;
  if ((timer_read('page') / 1000) > $max_exec_time) {
    return FALSE;
  }
  else return TRUE;
}

/**
 *  checks whether the ImageGrabber is enabled for this node or not.
 *
 *  @param $nid     node to detect
 *
 *  @return
 *          TRUE if ImageGrabber enabled else FALSE
 */
function feedapi_imagegrabber_is_enabled($nid) {
  $enabled = db_result(db_query("SELECT enabled FROM {feedapi_imagegrabber} WHERE nid = %d", $nid));
  return $enabled;
}

/**
 * returns the image field of the feed-item content type in which
 * ImageGrabber stores the image for the feed-item.
 *
 *  @param $nid node to select
 *
 *  @return
 *          the image field or else FALSE
 */
function feedapi_imagegrabber_image_field($nid) {

  $image_field = db_result(db_query("SELECT image_field FROM {feedapi_imagegrabber} WHERE nid = %d", $nid));
  return $image_field;
}

/**
 *  returns the ImageGrabber settings associated with a feed
 *
 *  @param $nid node to select
 *
 *  @return
 *          the settings array or else FALSE
 */
function feedapi_imagegrabber_get_settings($nid) {

  $settings = db_fetch_array(db_query("SELECT enabled, image_field, refresh_on_update, id_class, id_class_description, feeling_lucky FROM {feedapi_imagegrabber} WHERE nid = %d", $nid));
  return $settings;
}

/**
 * Processes the feed options for ImageGrabber and call the appropriate helper
 * function to download the image for the node.
 *
 * @param string $original_url url of the feed-item
 * @param integer $id_class whether tag identified by id or class or none
 * @param string $id_class_description the id or class of the tag
 * @param integer $feeling_lucky
 *
 * @return selected file on success, FALSE on failure
 */
function feedapi_imagegrabber_select_image($original_url, $id_class, $id_class_description, $feeling_lucky) {

  $filename = '';
  $options = array(
    // return web page
    CURLOPT_RETURNTRANSFER => TRUE,
    // fail on error
    CURLOPT_FAILONERROR => TRUE,
    // don't return headers
    CURLOPT_HEADER => FALSE,
    // follow redirects
    CURLOPT_FOLLOWLOCATION => TRUE,
    // handle all encodings
    CURLOPT_ENCODING => "",
    // who am i
    CURLOPT_USERAGENT => "feedapi_imagegrabber",
    // set referer on redirect
    CURLOPT_AUTOREFERER => TRUE,
    // timeout on connect
    CURLOPT_CONNECTTIMEOUT => 5,
    // timeout on response
    CURLOPT_TIMEOUT => 5,
    // stop after 10 redirects
    CURLOPT_MAXREDIRS => 2,
  );

  $ch = curl_init($original_url);
  curl_setopt_array($ch, $options);
  $file = curl_exec($ch);
  curl_close($ch);
  if ($file === FALSE) {
    return FALSE;
  }
  $doc = new DOMDocument();
  @$doc->loadHTML($file);

  if ($id_class == 0) {
    $items = $doc->getElementsByTagName("body");
    if ($items != NULL && $items->length > 0) {
      $dist = $items->item(0);
    }
    else $dist = NULL;
  }
  elseif ($id_class == 1) {
    $dist = $doc->getElementById($id_class_description);
  }
  elseif ($id_class == 2) {
    $xpath = new DOMXPath($doc);
    $items = $xpath->query("//*[@class and contains(concat(' ',normalize-space(@class),' '),' $id_class_description ')]");
    if ($items != NULL && $items->length > 0) {
      $dist = $items->item(0);
    }
    else $dist = NULL;
  }
  else {
    //Not Possible
    return FALSE;
  }
  if ($dist == NULL) {
    return FALSE;
  }

  $xml = $dist->ownerDocument->saveXML($dist);
  if ($feeling_lucky == 0) {
    $filename = feedapi_imagegrabber_search_images_to_select($original_url, $xml);
    return $filename;
  }
  else {
    $filename = feedapi_imagegrabber_download_image_feeling_lucky($original_url, $xml);
    return $filename;
  }
}

/**
 * Browse through all the images between the user defined tag to select the image.
 *
 * @param string $feed_item_url
 * @param string $xml
 *
 * @return filename on suceess, FALSE on failure.
 */
function feedapi_imagegrabber_search_images_to_select($feed_item_url, $xml) {

  $image_found   = 0;
  $selected_file = '';
  $selected_url  = '';
  $doc           = new DOMDocument();
  @$doc->loadXML($xml);
  $xpath            = new DOMXPath($doc);
  $hrefs            = $xpath->evaluate("//img");
  $temp_dir         = file_directory_temp();
  $size_image_found = 0;
  for ($i = 0; $i < $hrefs->length; $i++) {
    $href = $hrefs->item($i);
    $url = $href->getAttribute('src');
    if (!isset($url) || empty($url) || $url == '') {
      continue;
    }
    $abs = url_to_absolute($feed_item_url, $url);
    if ($abs != FALSE) {
      if (feedapi_imagegrabber_download_image_heuristics($abs, $size_image_found)) {
        $image_found = 1;
        $selected_url = $abs;
      }
    }
  }
  if ($image_found) {
    $selected_file = feedapi_imagegrabber_create_filename($temp_dir, $selected_url);
    if (!feedapi_imagegrabber_get_image($selected_url, $selected_file)) {
      return FALSE;
    }
    else return $selected_file;
  }
  else return FALSE;
}

/**
 * Heuristics for selecting the right image from the original feed item
 * of the URL when $feeling_lucky is FALSE
 *
 * 1) select the image greater than some minimum resolution.(100x150)
 * 2) select the largest amongst the images found.
 *
 * @param $image_url	url of the image.
 * @param $imagesize    size of the largest image found till now
 *
 * @return  boolean
 *          TRUE if image url was selected.
 *          FALSE if rejected or any other error as timeout, invalid url, etc.
 *
 */
function feedapi_imagegrabber_download_image_heuristics($image_url, &$imagesize) {

  if (!feedapi_imagegrabber_remote_image_exists($image_url)) {

    return FALSE;

  }

  $info = @getimagesize($image_url);

  if (isset($info) && is_array($info)) {

    // allowed drupal images
    $extensions = array('1' => 'gif', '2' => 'jpg', '3' => 'png', '4' => 'jpeg');
    if (array_key_exists($info[2], $extensions)) {
      $height     = $info[0];
      $width      = $info[1];
      $resolution = $height * $width;
      //Minimum resolution of 100x150
      if ($resolution < 15000 || $resolution < $imagesize) {
        return FALSE;
      }
      $imagesize = $resolution;
      return TRUE;
    }
  }
  return FALSE;
}

/**
 * Selects the first image between the user defined tag.
 *
 * @param string $feed_item_url
 * @param string $xml
 *
 * @return filename on suceess, FALSE on failure.
 */
function feedapi_imagegrabber_download_image_feeling_lucky($feed_item_url, $xml) {
  $temp_dir = file_directory_temp();
  $doc = new DOMDocument();
  @$doc->loadXML($xml);
  $xpath = new DOMXPath($doc);
  $hrefs = $xpath->evaluate("//img");
  if ($hrefs->length <= 0) {
    return FALSE;
  }
  for ($i = 0; $i < $hrefs->length; $i++) {
    $href = $hrefs->item($i);
    $url = $href->getAttribute('src');
    if (!isset($url) || empty($url) || $url == '') {
      continue;
    }
    $abs = url_to_absolute($feed_item_url, $url);
    if ($abs != FALSE) {
      $image_file = feedapi_imagegrabber_create_filename($temp_dir, $abs);
      if (!feedapi_imagegrabber_get_image($abs, $image_file)) {
        continue;
      }
      else {
        return $image_file;
      }
    }
  }
  return FALSE;
}

/**
 * Downloads an image from the given URL and stores it into the given
 * filename.
 *
 * @param   $image_url  the url of the image
 * @param   $filename   the name of the file.
 *
 * @return boolean
 * true on sucess else false for timeout, invalid response error or if the
 * downloaded file is not a drupal recognized image.
 */
function feedapi_imagegrabber_get_image($image_url, $filename) {
  if (function_exists('curl_exec')) {
    $options = array(
      // return web page
      CURLOPT_RETURNTRANSFER => TRUE,
      // fail on error
      CURLOPT_FAILONERROR => TRUE,
      // don't return headers
      CURLOPT_HEADER => FALSE,
      // follow redirects
      CURLOPT_FOLLOWLOCATION => TRUE,
      // who am i
      CURLOPT_USERAGENT => "feedapi_imagegrabber",
      // set referer on redirect
      CURLOPT_AUTOREFERER => TRUE,
      // timeout on connect
      CURLOPT_CONNECTTIMEOUT => 5,
      // timeout on response
      CURLOPT_TIMEOUT => 5,
      // stop after 10 redirects
      CURLOPT_MAXREDIRS => 2,
    );
    $ch = curl_init($image_url);
    curl_setopt_array($ch, $options);
    $content = curl_exec($ch);
    $err = curl_errno($ch);
    curl_close($ch);
    if (!$err && (file_put_contents($filename, $content) !== FALSE)) {
      return feedapi_imagegrabber_is_image($filename);
    }
  }
  return FALSE;

}

/**
 * Checks whether is file is a valid drupal recognized image or not.
 *
 * @param $filepath
 *   Path to the file to be checked
 *
 * @return boolean
 */
function feedapi_imagegrabber_is_image($filepath) {

  $info = image_get_info($filepath);
  if (!$info || empty($info['extension'])) {
    return FALSE;
  }
  return TRUE;
}

/**
 * creates a unique filename for the file,
 * Look at [#554980] for more
 *
 * @param string $dir the directory in which file is to created
 * @param string $url the url of the filenme
 * @param integer $count the unique file count
 *
 * @return destination filepath
 */
function feedapi_imagegrabber_create_filename($dir, $url) {
  $count = time();
  $arr = parse_url($url);
  if ($arr != FALSE) {
    $filename = basename($arr['path']);
    if (isset($filename) && !empty($filename) && $filename != '') {
      $filename_parts = explode('.', $filename);
      // Remove file basename.
      $new_filename = array_shift($filename_parts);
      // Remove final extension.
      $final_extension = array_pop($filename_parts);
      $short_filename = drupal_strlen($new_filename) > 100 ? drupal_substr($new_filename, 0, 100) : $new_filename;
      return file_destination(file_create_path($dir .'/'. $short_filename .'_'. $count .'.'. $final_extension), FILE_EXISTS_RENAME);
    }
  }
  return file_destination(file_create_path($dir .'/feedapi_imagegrabber_'. $count), FILE_EXISTS_RENAME);
}

/**
 * page callback for hook_menu().
 */
function feedapi_imagegrabber_download_images($node) {
  return drupal_get_form('feedapi_imagegrabber_download_confirm', $node);
}

/**
 * confirms the downloading of the images from where it was paused
 */
function feedapi_imagegrabber_download_confirm($form_state, $node) {
  $output = confirm_form(
    array('nid' => array('#type' => 'hidden', '#value' => $node->nid)),
    t('Grab images for the feed-items of !name', array('!name' => $node->title)), isset($_GET['destination']) ? $_GET['destination'] : 'node/'. $node->nid,
    t("Due to PHP execution time limit, some feed-items might have been left by ImageGrabber during last refresh. Are you sure you want to proceed with the download of those feed-items?"),
    t('Yes'), t('No'),
    'feedapi_imagegrabber_download_confirm'
  );
  return $output;
}

/**
 * submit for the grab images form.
 */
function feedapi_imagegrabber_download_confirm_submit($form, &$form_state) {

  $nid = $form_state['values']['nid'];
  $data = db_fetch_array(db_query("SELECT feed_nid, nid FROM {feedapi_imagegrabber_cron_nodes} WHERE feed_nid = %d", $nid));
  if ($data == FALSE) {
    drupal_set_message(t('No feed-items available for downloading images.'));
    $form_state['redirect'] = 'node/'. $form_state['values']['nid'];
    return;
  }
  $feed_items = unserialize($data['nid']);
  $feed = feedapi_imagegrabber_create_feed($nid, $feed_items);
  feedapi_imagegrabber_feedapi_after_refresh($feed, TRUE);
  $form_state['redirect'] = 'node/'. $form_state['values']['nid'];
}

/**
 * Creates a feed object from a node => url array.
 *
 * @param int $feed_nid the node of the feed
 * @param mixed $array array of feed items
 *
 * @return $feed, as created by feedapi
 */
function feedapi_imagegrabber_create_feed($feed_nid, $array) {
  $feed        = new stdClass();
  $feed->nid   = $feed_nid;
  $feed->items = new stdClass();
  $item        = 0;
  foreach ($array as $node => $url) {
    $temp_item = new stdClass();
    $temp_item->nid = $node;
    $options = new stdClass();
    $options->original_url = $url;
    $temp_item->options = $options;
    $feed->items->$item = $temp_item;
    $item++;
  }
  return $feed;
}

/**
 * Checks if a remote file exists and is below some specific size (1 MB)
 *
 * @param $url
 *
 * @return boolean
 */
function feedapi_imagegrabber_remote_image_exists($url) {

  $ch = curl_init();
  curl_setopt($ch, CURLOPT_URL, $url);
  curl_setopt($ch, CURLOPT_HEADER, 1);
  curl_setopt($ch, CURLOPT_NOBODY, 1);
  curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
  curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 5);
  curl_setopt($ch, CURLOPT_TIMEOUT, 5);
  curl_setopt($ch, CURLOPT_FAILONERROR, 1);
  curl_setopt($ch, CURLOPT_USERAGENT, "feedapi_imagegraber");
  curl_setopt($ch, CURLOPT_FOLLOWLOCATION, TRUE);
  curl_setopt($ch, CURLOPT_MAXREDIRS, 2);

  $result = curl_exec($ch);
  $ret = FALSE;
  if ($result !== FALSE) {
    // Who keeps such a big image expect a hacker (cracker)
    $max_imagesize  = 1024 * 1024;
    $statusCode     = curl_getinfo($ch, CURLINFO_HTTP_CODE);
    $content_length = curl_getinfo($ch, CURLINFO_CONTENT_LENGTH_DOWNLOAD);
    if ($statusCode == 200 && is_numeric($content_length) && $content_length < $max_imagesize) {
      $ret = TRUE;
    }
  }
  curl_close($ch);
  return $ret;
}