Feature #16534 » indexed_search_4167_v1.diff
/home/mario/eclipse_workspace/TYPO3/typo3/sysext/indexed_search/class.crawler.php (working copy) | ||
---|---|---|
27 | 27 |
/** |
28 | 28 |
* Crawler hook for indexed search. Works with the "crawler" extension |
29 | 29 |
* |
30 |
* @author Kasper Sk�rh�j <kasperYYYY@typo3.com>
|
|
30 |
* @author Kasper Skaarhoj <kasperYYYY@typo3.com>
|
|
31 | 31 |
*/ |
32 | 32 |
/** |
33 | 33 |
* [CLASS/FUNCTION INDEX of SCRIPT] |
... | ... | |
573 | 573 |
* |
574 | 574 |
* @param string URL string to check |
575 | 575 |
* @param array Array of already indexed URLs (input url is looked up here and must not exist already) |
576 |
* @param string Base URL of the indexing process (input URL must be "inside" the base URL!) |
|
576 |
* @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.
|
|
577 | 577 |
* @return string Returls the URL if OK, otherwise false |
578 | 578 |
*/ |
579 | 579 |
function checkUrl($url,$urlLog,$baseUrl) { |
... | ... | |
578 | 578 |
*/ |
579 | 579 |
function checkUrl($url,$urlLog,$baseUrl) { |
580 | 580 |
$url = ereg_replace('\/\/$','/',$url); |
581 | ||
582 |
// just geht the root of the URL like http://www.domain.tld/ |
|
583 |
$baseUrlArray = parse_url($baseUrl); |
|
584 |
$baseUrl = $baseUrlArray['scheme'] . '://'. $baseUrlArray['host'] .'/'; |
|
585 | ||
581 | 586 |
list($url) = explode('#',$url); |
582 | 587 | |
583 | 588 |
if (!strstr($url,'../')) { |