Feature #13680 » linkval-cleanup.diff
| classes/linktype/class.tx_linkvalidator_linktype_external.php (Arbeitskopie) | ||
|---|---|---|
|
protected $urlErrorParams = array();
|
||
|
/**
|
||
|
* List of headers to be used for metching an URL for the current processing
|
||
|
*
|
||
|
* @var array
|
||
|
*/
|
||
|
protected $additionalHeaders = array();
|
||
|
/**
|
||
|
* Checks a given URL + /path/filename.ext for validity
|
||
|
*
|
||
|
* @param string $url: url to check
|
||
| ... | ... | |
|
*/
|
||
|
public function checkLink($url, $softRefEntry, $reference) {
|
||
|
$errorParams = array();
|
||
|
$report = array();
|
||
|
$additionalHeaders['User-Agent'] = 'User-Agent: Mozilla/5.0 TYPO3-linkvalidator';
|
||
|
if (isset($this->urlReports[$url])) {
|
||
|
if(!$this->urlReports[$url]) {
|
||
|
if(is_array($this->urlErrorParams[$url])) {
|
||
| ... | ... | |
|
}
|
||
|
// try to fetch the content of the URL (headers only)
|
||
|
$report = array();
|
||
|
$content = t3lib_div::getURL($url, 2, $additionalHeaders, $report);
|
||
|
// try fetching the content of the URL (just fetching the headers does not work correctly)
|
||
|
$content = '';
|
||
|
$content = t3lib_div::getURL($url, 1, FALSE, $report);
|
||
|
$tries = 0;
|
||
|
$currentUrl = $url;
|
||
|
$haveNewUrl = FALSE;
|
||
|
$location = array();
|
||
|
$cookie = array();
|
||
|
while (($report['http_code'] == 301 || $report['http_code'] == 302
|
||
|
|| $report['http_code'] == 303 || $report['http_code'] == 307)
|
||
|
&& ($tries < 5)) {
|
||
|
$isCodeRedirect = preg_match('/Location: (.*)/', $content, $location);
|
||
|
if (isset($location[1])) {
|
||
|
$content = t3lib_div::getURL($location[1], 2, FALSE, $report);
|
||
|
// split header into lines and parse them
|
||
|
$responseHeaders = t3lib_div::trimExplode(chr(10), $content, TRUE);
|
||
|
$cookies = '';
|
||
|
foreach($responseHeaders as $line) {
|
||
|
if (preg_match('/Location: ([^\r\n]+)/', $line, $location)) {
|
||
|
if (isset($location[1])) {
|
||
|
$parsedUrl = parse_url($location[1]);
|
||
|
if (!isset($parsedUrl['host'])) {
|
||
|
// the location did not contain a complete URI, build it!
|
||
|
$parsedUrl = parse_url($lastUrl);
|
||
|
$newUrl = $parsedUrl['scheme'].'://'.(isset($parsedUrl['user'])?$parsedUrl['user'].(isset($parsedUrl['pass'])?':'.$parsedUrl['pass']:''):'').$parsedUrl['host'].(isset($parsedUrl['port'])?':'.$parsedUrl['port']:'').$location[1];
|
||
|
}
|
||
|
else {
|
||
|
$newUrl = $location[1];
|
||
|
}
|
||
|
if ($currentUrl === $newUrl) break 2;
|
||
|
$currentUrl = $newUrl;
|
||
|
$haveNewUrl = TRUE;
|
||
|
} else break 2;
|
||
|
}
|
||
|
// find cookies
|
||
|
if (preg_match('/Set-Cookie: (.*); path=(.+)/', $line, $cookie)) {
|
||
|
$cookies .= (isset($cookie[1]) ? $cookie[1] : '') . '; ';
|
||
|
}
|
||
|
}
|
||
|
// add cookies if we have some
|
||
|
if (!empty($cookies)) $additionalHeaders['Cookie'] = 'Cookie: ' . $cookies;
|
||
|
// now try to fetch again
|
||
|
if ($haveNewUrl) {
|
||
|
$content = t3lib_div::getURL($currentUrl, 2, $additionalHeaders, $report);
|
||
|
$tries++;
|
||
|
} else break;
|
||
|
}
|
||
|
|
||
|
$response = TRUE;
|
||
|
// analyze the response
|
||
| ... | ... | |
|
if (($report['http_code'] == 301) || ($report['http_code'] == 302)
|
||
|
|| ($report['http_code'] == 303) || ($report['http_code'] == 307)) {
|
||
|
$errorParams['errorType'] = $report['http_code'];
|
||
|
$errorParams['location'] = $location[1];
|
||
|
$errorParams['location'] = $currentUrl;
|
||
|
$response = FALSE;
|
||
|
}
|
||
|
if ($report['http_code'] == 404 || $report['http_code'] == 403) {
|
||
|
$errorParams['errorType'] = $report['http_code'];
|
||
|
$response = FALSE;
|
||
|
}
|
||
|
if ($report['http_code'] >= 300 && $response) {
|
||
|
$errorParams['errorType'] = $report['http_code'];
|
||
|
$response = FALSE;
|
||
| ... | ... | |
|
$response = sprintf($GLOBALS['LANG']->getLL('list.report.redirectloop'), $errorType, $errorParams['location']);
|
||
|
break;
|
||
|
case 403:
|
||
|
$response = $GLOBALS['LANG']->getLL('list.report.pageforbidden403');
|
||
|
break;
|
||
|
case 404:
|
||
|
$response = $GLOBALS['LANG']->getLL('list.report.pagenotfound404');
|
||
|
break;
|
||
|
case 403:
|
||
|
$response = $GLOBALS['LANG']->getLL('list.report.pageforbidden403');
|
||
|
break;
|
||
|
case 500:
|
||
|
$response = $GLOBALS['LANG']->getLL('list.report.internalerror500');
|
||
|
break;
|
||