pathologic.module in Pathologic 6.2

Pathologic text filter for Drupal.

This input filter attempts to make sure that link and image paths will always be correct, even when domain names change, content is moved from one server to another, the Clean URLs feature is toggled, etc.

File

pathologic.module

<?php

/**
 * @file
 * Pathologic text filter for Drupal.
 *
 * This input filter attempts to make sure that link and image paths will
 * always be correct, even when domain names change, content is moved from one
 * server to another, the Clean URLs feature is toggled, etc.
 */

/**
 * Implementation of hook_filter().
 */
function pathologic_filter($op, $delta = 0, $format = -1, $text = '') {
  if ($op === 'process' && $text !== '') {
    if (variable_get('filter_pathologic_href_' . $format, TRUE)) {

      // Do transformation on href paths.
      // Make relative.
      $text = preg_replace_callback(_pathologic_abs_regex('href', $format), '_pathologic_abs_to_rel', $text);

      // Transform attributes.
      // The "internal:" bit in the regex makes life easier for those migrating
      // from the Path Filter module.
      $text = preg_replace_callback('|href="(internal:)?([^/#][^:"#\\?]+)([\\?]([^#]+))?(#([^:"]*))?"|', '_pathologic_do_href', $text);
    }
    if (variable_get('filter_pathologic_src_' . $format, TRUE)) {

      // Do transformation on src paths.
      // Make relative.
      $text = preg_replace_callback(_pathologic_abs_regex('src', $format), '_pathologic_abs_to_rel', $text);

      // Transform attributes.
      // The "internal:" bit in the regex makes life easier for those migrating
      // from the Path Filter module.
      $text = preg_replace_callback('|src="(internal:)?([^/][^:"\\?]+)([\\?]([^#]+))?"|', '_pathologic_do_src', $text);
    }
    return $text;
  }
  elseif ($op === 'list') {
    return array(
      t('Pathologic'),
    );
  }
  elseif ($op === 'description') {
    return t('Corrects paths in links and images in your Drupal content in situations which would otherwise cause them to &ldquo;break.&rdquo;');
  }
  elseif ($op === 'settings') {
    return array(
      'filter_pathologic' => array(
        '#type' => 'fieldset',
        '#title' => t('Pathologic'),
        '#collapsible' => TRUE,
        'filter_pathologic_href_' . $format => array(
          '#type' => 'checkbox',
          '#title' => t('Transform values of <em>href</em> attributes'),
          '#default_value' => intval(variable_get('filter_pathologic_href_' . $format, 1)),
          '#description' => t('<em>href</em> attributes are used in link tags.'),
        ),
        'filter_pathologic_src_' . $format => array(
          '#type' => 'checkbox',
          '#title' => t('Transform values of <em>src</em> attributes'),
          '#default_value' => intval(variable_get('filter_pathologic_src_' . $format, 1)),
          '#description' => t('<em>src</em> attributes are used in image tags and tags for other types of embedded media.'),
        ),
        'filter_pathologic_abs_paths_' . $format => array(
          '#type' => 'textarea',
          '#title' => t('Additional paths to be considered local'),
          '#default_value' => variable_get('filter_pathologic_abs_paths_' . $format, ''),
          '#description' => t('Enter URIs of other Drupal installations which should be considered local in addition to the one for this particular Drupal installation (which is %path). If in doubt, enter the path to the Drupal installation&rsquo;s front page. Enter one path per line.', array(
            '%path' => _pathologic_url('<front>'),
          )),
        ),
      ),
    );
  }
  return $text;
}

/**
 * Return the hard part of the regular expression to be used when making paths
 * relative.
 *
 * @param $attr
 *   The attribute holding paths we're looking for. Should be 'href' or 'src'.
 * @param $format
 *   The ID of the input format being used.
 * @return string
 *   Part of a regular expression to be used to find paths to act upon.
 */
function _pathologic_abs_regex($attr, $format) {
  static $pathstr = FALSE;
  if ($pathstr === FALSE) {
    $paths_field = trim(variable_get('filter_pathologic_abs_paths_' . $format, ''));
    if ($paths_field !== '') {

      // Get rid of spirious white space on each line.
      $paths = array_map('trim', explode("\n", $paths_field));
    }
    else {
      $paths = array();
    }
    $paths[] = _pathologic_url('<front>');

    // It's possible the user entered the path for the current site in the box,
    // especially if the DB contents are shared between two different servers
    // (likely if it's a testing/production server thing). This won't screw up
    // the regex, but it will unnecessarily complicate it, so let's remove
    // duplicates from the array.
    $paths = array_unique($paths);

    // We need to account for the fact that http://example.com/foo,
    // http://example.com/?q=foo and http://example.com/index.php?q=foo are all
    // valid.
    $pathstr = ')="(' . implode('/?(index.php)?(\\?q=)?|', $paths) . '/?(index.php)?(\\?q=)?)([^"]*)"`';

    // $pathstr now looks like:
    // )="(http://example.com//?(index.php)?(\?q=)?|http://example.org//?(index.php)?(\?q=)?)([^"]*)";
  }
  return '`(' . $attr . $pathstr;

  // Returned value looks like:
  // (href)="(http://example.com//?(index.php)?(\?q=)?|http://example.org//?(index.php)?(\?q=)?)([^"]*)"
  // We want to match the attribute so that the callback
  // _pathologic_abs_to_rel() below can return a value with the attribute
  // without special trickery or creating duplicate functions.
}

/**
 * Aid in turning an absolute path on the local server to a relative one.
 *
 * preg_replace_callback() callback function.
 */
function _pathologic_abs_to_rel($matches) {

  // The attribute will be the first match (after 0), and the path will be the
  // last.
  $path = array_pop($matches);
  if ($path === '') {

    // It looks like this was meant to be a link to the front page.
    return $matches[1] . '="<front>"';
  }
  return $matches[1] . '="' . $path . '"';
}

/**
 * Properly format an HREF element.
 *
 * Here's where the magic happens. preg_replace_callback() callback function.
 */
function _pathologic_do_href($matches) {
  return 'href="' . _pathologic_url($matches[2], isset($matches[4]) ? $matches[4] : NULL, isset($matches[6]) ? $matches[6] : NULL) . '"';
}

/**
 * Return a formatted SRC attribute.
 *
 * You might think SRC attributes will always be disk files instead of Drupal
 * paths, but that's not always the case - paths to files processed by
 * Imagecache point to non-existant files the first time, at which time they go
 * through the standard menu router to be caught and handled by Imagecache.
 * preg_replace_callback() callback function.
 */
function _pathologic_do_src($matches) {
  if (file_exists($matches[2])) {
    global $base_url;
    return "src=\"{$base_url}/{$matches[2]}\"";
  }
  return 'src="' . _pathologic_url($matches[2], isset($matches[4]) ? $matches[4] : NULL) . '"';
}

/**
 * Run the path parts through url() to puild a correct URL if the path looks
 * like it links to a local Drupal path.
 *
 * Otherwise, it's a link to a file or something -- try to return it in one
 * piece.
 *
 * @param $path
 *   The path to check.
 * @param $query
 *   The query fragment of the path, if any (comes after ?).
 * @param $fragment
 *   The anchor fragment of the path, if any (comes after #).
 * @return string
 *   The corrected path, if necessary. Else, the reconstructed path.
 */
function _pathologic_url($path, $query = NULL, $fragment = NULL) {

  // We're unescaping HTML entities like &amp; in the hope of avoiding double-
  // encoding. See: http://drupal.org/node/665900
  $path = html_entity_decode($path, ENT_COMPAT, 'UTF-8');

  // Does this look like an internal URL?
  if ($path !== '<front>' && !menu_get_item($path) && !drupal_lookup_path('source', $path)) {

    // It's not.
    global $base_url;
    $return = $base_url . '/' . $path;
    if ($query !== NULL && $query !== '') {
      $return .= '?' . $query;
    }
    if ($fragment !== NULL && $fragment !== '') {
      $return .= '#' . $fragment;
    }
    return $return;
  }
  return url($path, array(
    'query' => $query,
    'fragment' => $fragment,
    'absolute' => TRUE,
  ));
}

Functions

Name	Description
pathologic_filter	Implementation of hook_filter().
_pathologic_abs_regex	Return the hard part of the regular expression to be used when making paths relative.
_pathologic_abs_to_rel	Aid in turning an absolute path on the local server to a relative one.
_pathologic_do_href	Properly format an HREF element.
_pathologic_do_src	Return a formatted SRC attribute.
_pathologic_url	Run the path parts through url() to puild a correct URL if the path looks like it links to a local Drupal path.