function _pathologic_replace in Pathologic 7.3
Same name and namespace in other branches
- 8 pathologic.module \_pathologic_replace()
- 6.3 pathologic.module \_pathologic_replace()
- 7 pathologic.module \_pathologic_replace()
- 7.2 pathologic.module \_pathologic_replace()
Process and replace paths. preg_replace_callback() callback.
1 string reference to '_pathologic_replace'
- _pathologic_filter in ./
pathologic.module - Pathologic filter callback.
File
- ./
pathologic.module, line 213 - Pathologic text filter for Drupal.
Code
function _pathologic_replace($matches) {
// Get the base path.
global $base_path;
// Get the settings for the filter. Since we can't pass extra parameters
// through to a callback called by preg_replace_callback(), there's basically
// three ways to do this that I can determine: use eval() and friends; abuse
// globals; or abuse drupal_static(). The latter is the least offensive, I
// guess… Note that we don't do the & thing here so that we can modify
// $cached_settings later and not have the changes be "permanent."
$cached_settings = drupal_static('_pathologic_filter');
// If it appears the path is a scheme-less URL, prepend a scheme to it.
// parse_url() cannot properly parse scheme-less URLs. Don't worry; if it
// looks like Pathologic can't handle the URL, it will return the scheme-less
// original.
// @see https://drupal.org/node/1617944
// @see https://drupal.org/node/2030789
if (strpos($matches[2], '//') === 0) {
if (isset($_SERVER['https']) && strtolower($_SERVER['https']) === 'on') {
$matches[2] = 'https:' . $matches[2];
}
else {
$matches[2] = 'http:' . $matches[2];
}
}
// Now parse the URL after reverting HTML character encoding.
// @see http://drupal.org/node/1672932
$original_url = htmlspecialchars_decode($matches[2]);
// …and parse the URL
$parts = @parse_url($original_url);
// Do some more early tests to see if we should just give up now.
if (empty($parts) || isset($parts['scheme']) && !in_array($parts['scheme'], variable_get('pathologic_scheme_whitelist', array(
'http',
'https',
'files',
'internal',
))) || isset($parts['fragment']) && count($parts) === 1) {
// Give up by "replacing" the original with the same.
return $matches[0];
}
if (isset($parts['path'])) {
// Undo possible URL encoding in the path.
// @see http://drupal.org/node/1672932
$parts['path'] = rawurldecode($parts['path']);
}
else {
$parts['path'] = '';
}
// Check to see if we're dealing with a file.
// @todo Should we still try to do path correction on these files too?
if (isset($parts['scheme']) && $parts['scheme'] === 'files') {
// Path Filter "files:" support. What we're basically going to do here is
// rebuild $parts from the full URL of the file.
$new_parts = @parse_url(file_create_url(file_default_scheme() . '://' . $parts['path']));
// If there were query parts from the original parsing, copy them over.
if (!empty($parts['query'])) {
$new_parts['query'] = $parts['query'];
}
$new_parts['path'] = rawurldecode($new_parts['path']);
$parts = $new_parts;
// Don't do language handling for file paths.
$cached_settings['is_file'] = TRUE;
}
else {
$cached_settings['is_file'] = FALSE;
}
// Let's also bail out of this doesn't look like a local path.
$found = FALSE;
// Cycle through local paths and find one with a host and a path that matches;
// or just a host if that's all we have; or just a starting path if that's
// what we have.
foreach ($cached_settings['current_settings']['local_paths_exploded'] as $exploded) {
// If a path is available in both…
if (isset($exploded['path']) && isset($parts['path']) && strpos($parts['path'], $exploded['path']) === 0 && (isset($exploded['host']) && isset($parts['host']) && $exploded['host'] === $parts['host'] || !isset($exploded['host']) && !isset($parts['host']))) {
// Remove the shared path from the path. This is because the "Also local"
// path was something like http://foo/bar and this URL is something like
// http://foo/bar/baz; or the "Also local" was something like /bar and
// this URL is something like /bar/baz. And we only care about the /baz
// part.
$parts['path'] = drupal_substr($parts['path'], drupal_strlen($exploded['path']));
$found = TRUE;
// Break out of the foreach loop
break;
}
elseif (isset($parts['host']) && !isset($exploded['path']) && isset($exploded['host']) && $exploded['host'] === $parts['host']) {
// No further editing; just continue
$found = TRUE;
// Break out of foreach loop
break;
}
elseif (!isset($parts['host']) && (!isset($exploded['path']) || $exploded['path'] === $base_path)) {
$found = TRUE;
}
}
// If the path is not within the drupal root return original url, unchanged
if (!$found) {
return $matches[0];
}
// Okay, format the URL.
// If there's still a slash lingering at the start of the path, chop it off.
$parts['path'] = ltrim($parts['path'], '/');
// Examine the query part of the URL. Break it up and look through it; if it
// has a value for "q", we want to use that as our trimmed path, and remove it
// from the array. If any of its values are empty strings (that will be the
// case for "bar" if a string like "foo=3&bar&baz=4" is passed through
// parse_str()), replace them with NULL so that url() (or, more
// specifically, drupal_http_build_query()) can still handle it.
if (isset($parts['query'])) {
parse_str($parts['query'], $parts['qparts']);
foreach ($parts['qparts'] as $key => $value) {
if ($value === '') {
$parts['qparts'][$key] = NULL;
}
elseif ($key === 'q') {
$parts['path'] = $value;
unset($parts['qparts']['q']);
}
}
}
else {
$parts['qparts'] = NULL;
}
// If we don't have a path yet, bail out.
if (!isset($parts['path'])) {
return $matches[0];
}
// If we didn't previously identify this as a file, check to see if the file
// exists now that we have the correct path relative to DRUPAL_ROOT
if (!$cached_settings['is_file']) {
$cached_settings['is_file'] = !empty($parts['path']) && (is_file(DRUPAL_ROOT . '/' . $parts['path']) || _pathologic_is_file_directory($parts['path']));
}
// Okay, deal with language stuff.
$language_list = language_list();
if ($cached_settings['is_file']) {
// If we're linking to a file, use a fake LANGUAGE_NONE language object.
// Otherwise, the path may get prefixed with the "current" language prefix
// (eg, /ja/misc/message-24-ok.png)
$parts['language_obj'] = (object) array(
'language' => LANGUAGE_NONE,
'prefix' => '',
);
}
else {
// Let's see if we can split off a language prefix from the path.
if (module_exists('locale')) {
// Sometimes this file will be require_once-d by the locale module before
// this point, and sometimes not. We require_once it ourselves to be sure.
require_once DRUPAL_ROOT . '/includes/language.inc';
list($language_obj, $path) = language_url_split_prefix($parts['path'], $language_list);
if ($language_obj) {
$parts['path'] = $path;
$parts['language_obj'] = $language_obj;
}
}
if (empty($parts['language_obj']) && !empty($cached_settings['langcode']) && !empty($language_list[$cached_settings['langcode']])) {
$parts['language_obj'] = $language_list[$cached_settings['langcode']];
}
}
// If we get to this point and $parts['path'] is now an empty string (which
// will be the case if the path was originally just "/"), then we
// want to link to <front>.
if ($parts['path'] === '') {
$parts['path'] = '<front>';
}
// Build the parameters we will send to url()
$url_params = array(
'path' => $parts['path'],
'options' => array(
'query' => $parts['qparts'],
'fragment' => isset($parts['fragment']) ? $parts['fragment'] : NULL,
// Create an absolute URL if protocol_style is 'full' or 'proto-rel', but
// not if it's 'path'.
'absolute' => $cached_settings['current_settings']['protocol_style'] !== 'path',
// If we seem to have found a language for the path, pass it along to
// url(). Otherwise, ignore the 'language' parameter.
'language' => isset($parts['language_obj']) ? $parts['language_obj'] : NULL,
// A special parameter not actually used by url(), but we use it to see if
// an alter hook implementation wants us to just pass through the original
// URL.
'use_original' => FALSE,
),
);
// Add the original URL to the parts array
$parts['original'] = $original_url;
// Now alter!
// @see http://drupal.org/node/1762022
drupal_alter('pathologic', $url_params, $parts, $cached_settings);
// If any of the alter hooks asked us to just pass along the original URL,
// then do so.
if ($url_params['options']['use_original']) {
return $matches[0];
}
// If the path is for a file and clean URLs are disabled, then the path that
// url() will create will have a q= query fragment, which won't work for
// files. To avoid that, we use this trick to temporarily turn clean URLs on.
// This is horrible, but it seems to be the sanest way to do this.
// @see http://drupal.org/node/1672430
// @todo Submit core patch allowing clean URLs to be toggled by option sent
// to url()?
if (!empty($cached_settings['is_file'])) {
$cached_settings['orig_clean_url'] = !empty($GLOBALS['conf']['clean_url']);
if (!$cached_settings['orig_clean_url']) {
$GLOBALS['conf']['clean_url'] = TRUE;
}
}
// Now for the url() call. Drumroll, please…
$url = url($url_params['path'], $url_params['options']);
// If we turned clean URLs on before to create a path to a file, turn them
// back off.
if ($cached_settings['is_file'] && !$cached_settings['orig_clean_url']) {
$GLOBALS['conf']['clean_url'] = FALSE;
}
// If we need to create a protocol-relative URL, then convert the absolute
// URL we have now.
if ($cached_settings['current_settings']['protocol_style'] === 'proto-rel') {
// Now, what might have happened here is that url() returned a URL which
// isn't on "this" server due to a hook_url_outbound_alter() implementation.
// We don't want to convert the URL in that case. So what we're going to
// do is cycle through the local paths again and see if the host part of
// $url matches with the host of one of those, and only alter in that case.
$url_parts = @parse_url($url);
if (!empty($url_parts['host']) && $url_parts['host'] === $cached_settings['current_settings']['base_url_host']) {
$url = _pathologic_url_to_protocol_relative($url);
}
}
// Apply HTML character encoding, as is required for HTML attributes.
// @see http://drupal.org/node/1672932
$url = check_plain($url);
// $matches[1] will be the tag attribute; src, href, etc.
return " {$matches[1]}=\"{$url}";
}