Bug #20770

[t3lib_div::getURL][curl] CURLOPT_HTTPAUTH missing

Added by Amar Bouchibane almost 4 years ago. Updated almost 2 years ago.

Status:Needs Feedback Start date:2009-07-19
Priority:Could have Due date:
Assignee:Amar Bouchibane % Done:

0%
Category:-
Target version:-
TYPO3 Version:4.6 Complexity:hard
PHP Version:5.2
Votes: 0

Description

When using a basic authentication (Apache) on site curl won't get the content because of missing CURLOPT_HTTPAUTH/CURLOPT_USERPWD setopts.

I have attached a patch of t3lib_div which requires a "curl" user with a password generated from data "curl" hashed with the "encryptionKey" (see getURL). I don't know if this is very handy.

(issue imported from #M11559)

class.t3lib_div.php (192.4 kB) Administrator Admin, 2009-07-19 22:55

Related issues

related to Core - Feature #28344: Include HTTP Request2 for better HTTP handling Closed 2011-07-20 2011-07-25

History

#1
Updated by Marcus Krause almost 4 years ago

The general idea is nice. However, username and password being hardcoded isn't handy at all and this patch won't make it to TYPO3 Core as such.

#2
Updated by Philipp Gampe almost 2 years ago

  • Target version deleted (0)

I think t3lib_http_request handles this as well.

This should be set to feedback for a month and then be closed.

#3
Updated by Thorsten Kahler almost 2 years ago

  • Status changed from New to Needs Feedback
  • Assignee set to Amar Bouchibane
  • Priority changed from Should have to Could have
  • TYPO3 Version changed from 4.2 to 4.6
  • Complexity set to hard

Hi Amar, could you please give some more details on this issue? Perhaps you could also provide the patch as unified diff?

#4
Updated by Amar Bouchibane almost 2 years ago

Hi Thorsten,

I had (and still have) turned into a portal where Apaches' Basic Authentication is activated. My site uses the "mnogosearch" extension (and thus the "pagepath" extension).

I can still reproduce the problem when I change and then save one record in the TYPO3 BE that has to be indexed.

The mnogosearch extension calls (via tx_mnogosearch_tcemain::processDatamap_afterDatabaseOperations) the tx_pagepath_api::getPagePath method. Within this method t3lib_div::getURL will be called. When I debug the code, I notice that the return value of getURL (= curl_exec) is always FALSE:

When I use the "$report" in getURL, I get back this array:

$report : array (
        error: 22,
        message: "The requested URL returned error: 401",
        lib: "CURL)

The two lines I've added (which worked then on a simple HTTP server without proxy) were these:

modified t3lib_div::getURL:
--------------------------
...
curl_setopt($ch, CURLOPT_FAILONERROR, 1);

// Added: Amar --- begin
curl_setopt($ch, CURLOPT_USERPWD, 'curl:'.hash_hmac('md5', 'curl', $GLOBALS['TYPO3_CONF_VARS']['SYS']['encryptionKey']));
curl_setopt($ch, CURLOPT_HTTPAUTH, CURLAUTH_ANY);
// Added: Amar --- end

// May fail (PHP 5.2.0 + and 5.1.5 +) when open_basedir or safe_mode are enabled
$followLocation = @curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
...

On the other hand the "Retrieve/Update" in the ExtManager although uses getURL(), but works perfectly... even with a "curlProxyServer".

Anyway my workaround doesn't work for me anymore (in TYPO3 4.4.7). I got to admit that I use SSL and a curl-Proxy now.
But it worked for me at that time with a simple HTTP-Server without a curl-Proxy.

Also available in: Atom PDF