pathologic.module in Pathologic 6

The main and only module file.

File

pathologic.module

<?php

/**
 * @defgroup pathologic 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, Clean URLs is toggled, etc.
 */

/**
 * @file
 * The main and only module file.
 *
 * @ingroup pathologic
 */

/**
 * 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
      $text = preg_replace_callback('|href="([^/#][^:"#]+)#?([^:"]*)"|', '_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
      $text = preg_replace_callback('|src="([^/][^:"]+)"|', '_pathologic_do_src', $text);
    }
    return $text;
  }
  if ($op === 'list') {
    return array(
      t('Pathologic'),
    );
  }
  if ($op === 'description') {
    return t('Corrects paths in content which reference pages or media on this server.');
  }
  if ($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. $attr will probably be either 'href' or 'src'.
 */
function _pathologic_abs_regex($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);
    $pathstr = ')="(' . implode('/?(index.php)?(\\?q=)?|', $paths) . '/?(index.php)?(\\?q=)?)([^"]*)"`';

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

  // Returned value looks like:
  // (href)="(http://abcde.fgh//?(index.php)?(\?q=)?|http://edcba.hgf//?(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.
}

/**
 * Aids in turning an absolute path to a page on the local server to a relative
 * path so that Pathologic can act on it further. 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 formats an HREF element. Here's where the magic happens…
 * preg_replace_callback() callback function.
 */
function _pathologic_do_href($matches) {
  return 'href="' . _pathologic_url($matches[1], $matches[2]) . '"';
}

/**
 * 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 - Imagecache relies on file paths
 * actually being Drupal paths, at least initially. preg_replace_callback()
 * callback function.
 */
function _pathologic_do_src($matches) {
  if (file_exists($matches[1])) {
    global $base_url;
    return "src=\"{$base_url}/{$matches[1]}\"";
  }
  return 'src="' . _pathologic_url($matches[1]) . '"';
}

/**
 * Run the parts through url() if they look like they link to a local Drupal
 * path. Otherwise, it's a link to a file or something -- try to return it
 * in one piece.
 */
function _pathologic_url($path, $fragment = NULL) {

  // 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 . '/' . $path;
    if ($fragment !== NULL && $fragment !== '') {
      $return .= '#' . $fragment;
    }
    return $return;
  }
  return url($path, array(
    'fragment' => $fragment,
    'absolute' => TRUE,
  ));
}

Related topics

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, Clean URLs is toggled, etc.

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. $attr will probably be either 'href' or 'src'.
_pathologic_abs_to_rel	Aids in turning an absolute path to a page on the local server to a relative path so that Pathologic can act on it further. preg_replace_callback() callback function.
_pathologic_do_href	Properly formats an HREF element. Here's where the magic happens… preg_replace_callback() callback function.
_pathologic_do_src	Return a formatted SRC attribute.
_pathologic_url	Run the parts through url() if they look like they link to a local Drupal path. Otherwise, it's a link to a file or something -- try to return it in one piece.