Feature #16534 » indexed_search_4167_v1.diff
| /home/mario/eclipse_workspace/TYPO3/typo3/sysext/indexed_search/class.crawler.php (working copy) | ||
|---|---|---|
|
/**
|
||
|
* Crawler hook for indexed search. Works with the "crawler" extension
|
||
|
*
|
||
|
* @author Kasper Sk�rh�j <kasperYYYY@typo3.com>
|
||
|
* @author Kasper Skaarhoj <kasperYYYY@typo3.com>
|
||
|
*/
|
||
|
/**
|
||
|
* [CLASS/FUNCTION INDEX of SCRIPT]
|
||
| ... | ... | |
|
*
|
||
|
* @param string URL string to check
|
||
|
* @param array Array of already indexed URLs (input url is looked up here and must not exist already)
|
||
|
* @param string Base URL of the indexing process (input URL must be "inside" the base URL!)
|
||
|
* @param string Base URL of the indexing process (input URL must be "inside" the base URL!). If the base URL is pointing to a file, the path to the file is stripped of for checking.
|
||
|
* @return string Returls the URL if OK, otherwise false
|
||
|
*/
|
||
|
function checkUrl($url,$urlLog,$baseUrl) {
|
||
| ... | ... | |
|
*/
|
||
|
function checkUrl($url,$urlLog,$baseUrl) {
|
||
|
$url = ereg_replace('\/\/$','/',$url);
|
||
|
// just geht the root of the URL like http://www.domain.tld/
|
||
|
$baseUrlArray = parse_url($baseUrl);
|
||
|
$baseUrl = $baseUrlArray['scheme'] . '://'. $baseUrlArray['host'] .'/';
|
||
|
list($url) = explode('#',$url);
|
||
|
if (!strstr($url,'../')) {
|
||