media.filter.inc

<?php

/**
 * @file
 * Functions related to the WYSIWYG editor and the media input filter.
 *
 * @TODO: Rename this file?
 */

/**
 * Implements hook_wysiwyg_include_directory().
 */
function media_wysiwyg_include_directory($type) {
  switch ($type) {
    case 'plugins':
      return 'wysiwyg_plugins';
        break;
    }
}

/**
 * Filter callback for media markup filter.
 *
 * @TODO check for security probably pass text through filter_xss
 * @return unknown_type
 */
function media_filter($text) {
  $text = ' ' . $text . ' ';
  $text = preg_replace_callback("/\[\[.*?\]\]/s", 'media_token_to_markup', $text);

  return $text;
}

/**
 * Parses the contents of a CSS declaration block and returns a keyed array of property names and values.
 *
 * @param $declarations
 *   One or more CSS declarations delimited by a semicolon. The same as a CSS
 *   declaration block (see http://www.w3.org/TR/CSS21/syndata.html#rule-sets),
 *   but without the opening and closing curly braces. Also the same as the
 *   value of an inline HTML style attribute.
 *
 * @return
 *   A keyed array. The keys are CSS property names, and the values are CSS
 *   property values.
 */
function media_parse_css_declarations($declarations) {
  $properties = array();
  foreach (array_map('trim', explode(";", $declarations)) as $declaration) {
    if ($declaration != '') {
      list($name, $value) = array_map('trim', explode(':', $declaration, 2));
      $properties[strtolower($name)] = $value;
    }
  }
  return $properties;
}

/**
 * Replace callback to convert a media file tag into HTML markup.
 *
 * @param string $match
 *   Takes a match of tag code
 * @param boolean $wysiwyg
 *   Set to TRUE if called from within the WYSIWYG text area editor.
 * @return
 *   The HTML markup representation of the tag, or an empty string on failure.
 *
 * @see media_get_file_without_label()
 * @see hook_media_token_to_markup_alter()
 */
function media_token_to_markup($match, $wysiwyg = FALSE) {
  $settings = array();
  $match = str_replace("[[", "", $match);
  $match = str_replace("]]", "", $match);
  $tag = $match[0];

  try {
    if (!is_string($tag)) {
      throw new Exception('Unable to find matching tag');
    }

    $tag_info = drupal_json_decode($tag);

    if (!isset($tag_info['fid'])) {
      throw new Exception('No file Id');
    }

    // Ensure a valid view mode is being requested.
    if (!isset($tag_info['view_mode'])) {
      $tag_info['view_mode'] = media_variable_get('wysiwyg_default_view_mode');
    }
    elseif ($tag_info['view_mode'] != 'default') {
      $file_entity_info = entity_get_info('file');
      if (!in_array($tag_info['view_mode'], array_keys($file_entity_info['view modes']))) {
        // Media 1.x defined some old view modes that have been superseded by
        // more semantically named ones in File Entity. The media_update_7203()
        // function updates field settings that reference the old view modes,
        // but it's impractical to update all text content, so adjust
        // accordingly here.
        static $view_mode_updates = array(
          'media_preview' => 'preview',
          'media_small' => 'teaser',
          'media_large' => 'full',
        );
        if (isset($view_mode_updates[$tag_info['view_mode']])) {
          $tag_info['view_mode'] = $view_mode_updates[$tag_info['view_mode']];
        }
        else {
          throw new Exception('Invalid view mode');
        }
      }
    }


    $file = file_load($tag_info['fid']);
    if (!$file) {
      throw new Exception('Could not load media object');
    }
    $tag_info['file'] = $file;

    // The class attributes is a string, but drupal requires it to be
    // an array, so we fix it here.
    if (!empty($tag_info['attributes']['class'])) {
      $tag_info['attributes']['class'] = explode(" ", $tag_info['attributes']['class']);
    }

    // Track the fid of this file in the {media_filter_usage} table.
    media_filter_track_usage($file->fid);

    $attributes = is_array($tag_info['attributes']) ? $tag_info['attributes'] : array();
    $attribute_whitelist = media_variable_get('wysiwyg_allowed_attributes');
    $settings['attributes'] = array_intersect_key($attributes, array_flip($attribute_whitelist));

    if (!empty($tag_info['attributes']) && is_array($tag_info['attributes'])) {
      $attribute_whitelist = media_variable_get('wysiwyg_allowed_attributes');
      $settings['attributes'] = array_intersect_key($tag_info['attributes'], array_flip($attribute_whitelist));

      // Many media formatters will want to apply width and height independently
      // of the style attribute or the corresponding HTML attributes, so pull
      // these two out into top-level settings. Different WYSIWYG editors have
      // different behavior with respect to whether they store user-specified
      // dimensions in the HTML attributes or the style attribute, so check both.
      // Per http://www.w3.org/TR/html5/the-map-element.html#attr-dim-width, the
      // HTML attributes are merely hints: CSS takes precedence.
      if (isset($settings['attributes']['style'])) {
        $css_properties = media_parse_css_declarations($settings['attributes']['style']);
        foreach (array('width', 'height') as $dimension) {
          if (isset($css_properties[$dimension]) && substr($css_properties[$dimension], -2) == 'px') {
            $settings[$dimension] = substr($css_properties[$dimension], 0, -2);
          }
          elseif (isset($settings['attributes'][$dimension])) {
            $settings[$dimension] = $settings['attributes'][$dimension];
          }
        }
      }
    }
  }
  catch (Exception $e) {
    watchdog('media', 'Unable to render media from %tag. Error: %error', array('%tag' => $tag, '%error' => $e->getMessage()));
    return '';
  }

  if ($wysiwyg) {
    $settings['wysiwyg'] = $wysiwyg;
    // If sending markup to a WYSIWYG, we need to pass the file infomation so
    // that a inline macro can be generated when the WYSIWYG is detached.
    // The WYSIWYG plugin is expecting this information in the format of a
    // urlencoded JSON string stored in the data-file_info attribute of the
    // element.
    $element = media_get_file_without_label($file, $tag_info['view_mode'], $settings);
    $data = drupal_json_encode(array(
      'type' => 'media',
      'fid' => $file->fid,
      'view_mode' => $tag_info['view_mode'],
    ));
    $element['#attributes']['data-file_info'] = urlencode($data);
    $element['#attributes']['class'][] = 'media-element';
  }
  else {
    // Display the field elements.
    $element = array();
    $element['content']['file'] = media_get_file_without_label($file, $tag_info['view_mode'], $settings);
    // Overwrite or set the file #alt attribute if it has been set in this instance.
    if (!empty($element['content']['file']['#attributes']['alt'])) {
      $element['content']['file']['#alt'] = $element['content']['file']['#attributes']['alt'];
    }
    // Overwrite or set the file #title attribute if it has been set in this instance.
    if (!empty($element['content']['file']['#attributes']['title'])) {
      $element['content']['file']['#title'] = $element['content']['file']['#attributes']['title'];
    }
    field_attach_prepare_view('file', array($file->fid => $file), $tag_info['view_mode']);
    entity_prepare_view('file', array($file->fid => $file));
    $element['content'] += field_attach_view('file', $file, $tag_info['view_mode']);
    if (count(element_children($element['content'])) > 1) {
      // Add surrounding divs to group them together.
      // We dont want divs when there are no additional fields to allow them
      // to display inline.
      $element['content']['#prefix'] = '<div>';
      $element['content']['#suffix'] = '</div>';
    }
  }
  drupal_alter('media_token_to_markup', $element, $tag_info, $settings);
  return drupal_render($element);
}

/**
 * Builds a map of media tags in the element being rendered to their rendered HTML.
 *
 * The map is stored in JS, so we can transform them when the editor is being displayed.
 *
 * @param array $element
 */
function media_pre_render_text_format($element) {
  // filter_process_format() copies properties to the expanded 'value' child
  // element.
  if (!isset($element['format'])) {
    return $element;
  }

  $field = &$element['value'];
  $settings = array(
    'field' => $field['#id'],
  );

  $tagmap = _media_generate_tagMap($field['#value']);

  if (isset($tagmap)) {
    drupal_add_js(array('tagmap' => $tagmap), 'setting');
  }
  return $element;
}

/**
 * Generates an array of [inline tags] => <html> to be used in filter
 * replacement and to add the mapping to JS.
 * @param
 * The String containing text and html markup of textarea
 * @return
 * An associative array with tag code as key and html markup as the value.
 *
 * @see media_process_form()
 * @see media_token_to_markup()
 */
function _media_generate_tagMap($text) {
  // Making $tagmap static as this function is called many times and
  // adds duplicate markup for each tag code in Drupal.settings JS,
  // so in media_process_form it adds something like tagCode:<markup>,
  // <markup> and when we replace in attach see two duplicate images
  // for one tagCode. Making static would make function remember value
  // between function calls. Since media_process_form is multiple times
  // with same form, this function is also called multiple times.
  static $tagmap = array();
  preg_match_all("/\[\[.*?\]\]/s", $text, $matches, PREG_SET_ORDER);
  foreach ($matches as $match) {
    // We see if tagContent is already in $tagMap, if not we add it
    // to $tagmap.  If we return an empty array, we break embeddings of the same
    // media multiple times.
    if (empty($tagmap[$match[0]])) {
      // @TODO: Total HACK, but better than nothing.
      // We should find a better way of cleaning this up.
      if ($markup_for_media = media_token_to_markup($match, TRUE)) {
        $tagmap[$match[0]] = $markup_for_media;
      }
      else {
        $missing = file_create_url(drupal_get_path('module', 'media') . '/images/icons/default/image-x-generic.png');
        $tagmap[$match[0]] = '<div><img src="' . $missing . '" width="100px" height="100px"/></div>';
      }
    }
  }
  return $tagmap;
}

/**
 * Return a list of view modes allowed for a file embedded in the WYSIWYG.
 *
 * @param object $file
 *   A file entity.
 *
 * @return array
 *   An array of view modes that can be used on the file when embedded in the
 *   WYSIWYG.
 */
function media_get_wysiwyg_allowed_view_modes($file) {
  $enabled_view_modes = &drupal_static(__FUNCTION__, array());

  // @todo Add more caching for this.
  if (!isset($enabled_view_modes[$file->type])) {
    $enabled_view_modes[$file->type] = array();

    // Add the default view mode by default.
    $enabled_view_modes[$file->type]['default'] = array('label' => t('Default'), 'custom settings' => TRUE);

    $entity_info = entity_get_info('file');
    $view_mode_settings = field_view_mode_settings('file', $file->type);
    foreach ($entity_info['view modes'] as $view_mode => $view_mode_info) {
      // Do not show view modes that don't have their own settings and will
      // only fall back to the default view mode.
      if (empty($view_mode_settings[$view_mode]['custom_settings'])) {
        continue;
      }

      // Don't present the user with an option to choose a view mode in which
      // the file is hidden.
      $extra_fields = field_extra_fields_get_display('file', $file->type, $view_mode);
      if (empty($extra_fields['file']['visible'])) {
        continue;
      }

      // Add the view mode to the list of enabled view modes.
      $enabled_view_modes[$file->type][$view_mode] = $view_mode_info;
    }
  }

  $view_modes = $enabled_view_modes[$file->type];
  drupal_alter('media_wysiwyg_allowed_view_modes', $view_modes, $file);
  return $view_modes;
}

/**
 * Form callback used when embedding media.
 *
 * Allows the user to pick a format for their media file.
 * Can also have additional params depending on the media type.
 */
function media_format_form($form, $form_state, $file) {
  $form = array();
  $form['#media'] = $file;

  $view_modes = media_get_wysiwyg_allowed_view_modes($file);
  $formats = $options = array();
  foreach ($view_modes as $view_mode => $view_mode_info) {
    //@TODO: Display more verbose information about which formatter and what it does.
    $options[$view_mode] = $view_mode_info['label'];
    $element = media_get_file_without_label($file, $view_mode, array('wysiwyg' => TRUE));

    // Make a pretty name out of this.
    $formats[$view_mode] = drupal_render($element);
  }

  // Add the previews back into the form array so they can be altered.
  $form['#formats'] = &$formats;

  if (!count($formats)) {
    throw new Exception('Unable to continue, no available formats for displaying media.');
    return;
  }

  $default_view_mode = media_variable_get('wysiwyg_default_view_mode');
  if (!isset($formats[$default_view_mode])) {
    $default_view_mode = key($formats);
  }

  // Add the previews by reference so that they can easily be altered by
  // changing $form['#formats'].
  $settings['media']['formatFormFormats'] = &$formats;
  $form['#attached']['js'][] = array('data' => $settings, 'type' => 'setting');

  // Add the required libraries, JavaScript and CSS for the form.
  $form['#attached']['library'][] = array('media', 'media_base');
  $form['#attached']['library'][] = array('system', 'form');
  $form['#attached']['js'][] = drupal_get_path('module', 'media') . '/js/media.format_form.js';

  $form['heading'] = array(
    '#type' => 'markup',
    '#prefix' => '<h1 class="title">',
    '#suffix' => '</h1>',
    '#markup' => t('Embedding %filename', array('%filename' => $file->filename)),
  );

  $preview = media_get_thumbnail_preview($file);

  $form['preview'] = array(
    '#type' => 'markup',
    '#title' => check_plain(basename($file->uri)),
    '#markup' => drupal_render($preview),
  );

  // These will get passed on to WYSIWYG
  $form['options'] = array(
    '#type' => 'fieldset',
    '#title' => t('options'),
  );

  $form['options']['format'] = array(
    '#type' => 'select',
    '#title' => t('Current format is'),
    '#options' => $options,
    '#default_value' => $default_view_mode
  );

  if ($file->type === 'image') {
    $form['options']['alt'] = array(
      '#type' => 'textfield',
      '#title' => t('Alternate text'),
      '#description' => t('This text will be used by screen readers, search engines, or when the image cannot be loaded.'),
      '#default_value' => isset($file->field_file_image_alt_text['und'][0]['safe_value']) ? $file->field_file_image_alt_text['und'][0]['safe_value'] : '',
    );

    $form['options']['title'] = array(
      '#type' => 'textfield',
      '#title' => t('Title'),
      '#description' => t('The title is used as a tool tip when the user hovers the mouse over the image.'),
      '#default_value' => isset($file->field_file_image_title_text['und'][0]['safe_value']) ? $file->field_file_image_title_text['und'][0]['safe_value'] : '',
    );
  }

  // Similar to a form_alter, but we want this to run first so that media.types.inc
  // can add the fields specific to a given type (like alt tags on media).
  // If implemented as an alter, this might not happen, making other alters not
  // be able to work on those fields.
  // @TODO: We need to pass in existing values for those attributes.
  drupal_alter('media_format_form_prepare', $form, $form_state, $file);

  if (!element_children($form['options'])) {
    $form['options']['#attributes'] = array('style' => 'display:none');
  }

  return $form;
}

/**
 * Returns a drupal_render() array for just the file portion of a file entity.
 *
 * Optional custom settings can override how the file is displayed.
 */
function media_get_file_without_label($file, $view_mode, $settings = array()) {
  $file->override = $settings;

  $element = file_view_file($file, $view_mode);

  // The formatter invoked by file_view_file() can use $file->override to
  // customize the returned render array to match the requested settings. To
  // support simple formatters that don't do this, set the element attributes to
  // what was requested, but not if the formatter applied its own logic for
  // element attributes.
  if (!isset($element['#attributes']) && isset($settings['attributes'])) {
    $element['#attributes'] = $settings['attributes'];

    // While this function may be called for any file type, images are a common
    // use-case. theme_image() and theme_image_style() require the 'alt'
    // attribute to be passed separately from the 'attributes' array (see
    // http://drupal.org/node/999338). Until that's fixed, implement this
    // special-case logic. Image formatters using other theme functions are
    // responsible for their own 'alt' attribute handling. See
    // theme_media_formatter_large_icon() for an example.
    if (isset($settings['attributes']['alt']) && !isset($element['#alt']) && isset($element['#theme']) && in_array($element['#theme'], array('image', 'image_style'))) {
      $element['#alt'] = $settings['attributes']['alt'];
    }
  }

  return $element;
}

/**
 * Clears caches that may be affected by the media filter.
 *
 * The media filter calls file_load(). This means that if a file object
 * is updated, the check_markup() and field caches could return stale content.
 * There are several possible approaches to deal with this:
 *  - Disable filter caching in media_filter_info(), this was found to cause a
 *    30% performance hit from profiling four node teasers, due to both the
 *    media filter itself, and other filters that can't be cached.
 *  - Clear the filter and field caches whenever any media node is updated, this
 *    would ensure cache coherency but would reduce the effectiveness of those
 *    caches on high traffic sites with lots of media content updates.
 *  - The approach taken here: Record the fid of all media objects that are
 *    referenced by the media filter. Only clear the filter and field caches
 *    when one of these is updated, as opposed to all media objects.
 *  - @todo: consider an EntityFieldQuery to limit cache clearing to only those
 *    entities that use a text format with the media filter, possibly checking
 *    the contents of those fields to further limit this to fields referencing
 *    the media object being updated. This would need to be implemented
 *    carefully to avoid scalability issues with large result sets, and may
 *    not be worth the effort.
 *
 * @param $fid
 *   Optional media fid being updated. If not given, the cache will be cleared
 *   as long as any file is referenced.
 */
function media_filter_invalidate_caches($fid = FALSE) {
  // If fid is passed, confirm that it has previously been referenced by the
  // media filter. If not, clear the cache if the {media_filter_usage} has any
  // valid records.
  if (($fid && db_query('SELECT fid FROM {media_filter_usage} WHERE fid = :fid', array(':fid' => $fid))->fetchField()) || (!$fid && media_filter_usage_has_records())) {
    // @todo: support entity cache, either via a hook, or using module_exists().
    cache_clear_all('*', 'cache_filter', TRUE);
    cache_clear_all('*', 'cache_field', TRUE);
  }
}

/**
 * Determines if the {media_filter_usage} table has any entries.
 */
function media_filter_usage_has_records() {
  return  (bool) db_query_range('SELECT 1 FROM {media_filter_usage} WHERE fid > :fid', 0, 1, array(':fid' => 0))->fetchField();
}

/**
 * Tracks usage of media fids by the media filter.
 *
 * @param $fid
 *   The media fid.
 */
function media_filter_track_usage($fid) {
  // This function only tracks when fids are found by the media filter.
  // It would be impractical to check when formatted text is edited to remove
  // references to fids, however by keeping a timestamp, we can implement
  // rudimentary garbage collection in hook_flush_caches().
  // However we only need to track that an fid has ever been referenced,
  // not every time, so avoid updating this table more than once per month,
  // per fid.
  $timestamp = db_query('SELECT timestamp FROM {media_filter_usage} WHERE fid = :fid', array(':fid' => $fid))->fetchField();
  if (!$timestamp || $timestamp <= REQUEST_TIME - 86400 * 30) {
    db_merge('media_filter_usage')->key(array('fid' => $fid))->fields(array('fid' => $fid, 'timestamp' => REQUEST_TIME))->execute();
  }
}