HOWTO: Proper Links and URLs for XHTML

Copyright (c) 2007 Elmar Hinz, elmar.hinz@team-red.net
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts. A copy of the license is included in the section entitled “GNU Free Documentation License”.

How a link for XHTML should look like

The raw URL:

links.php?märker[einEuro]=1€&märker[fünfEuro]=5€

For XHTML we need to urlencode it, apart from the ampersand and the equal signs, which the PHP parser uses to analyse the variables. The ampersand signs must be used as entities.

links.php?m%C3%A4rker%5BeinEuro%5D=1%E2%82%AC&m%C3%A4rker%5Bf%C3%BCnfEuro%5D=5%E2%82%AC

The square brackets also can be encoded. But it also works without encoding them. According to RFC 1738 they should be encoded.

In PHP the function would be composed this way

$url = 'links.php?'
. rawurlencode('märker[einEuro]') . '=' . rawurlencode('1€')
. '&'
. rawurlencode('märker[fünfEuro]') . '=' . rawurlencode('5€');


An alternative way in 2 steps with htmlspecialchars

Step 1:

$url = 'links.php?'
. rawurlencode('märker[einEuro]') . '=' . rawurlencode('1€')
. '&'
. rawurlencode('märker[fünfEuro]') . '=' . rawurlencode('5€');

Step 2:

$url = htmlspecialchars($url);

This is possible due to the previous urlencoding of keys and values. No characters are remaining that would be altered by the htmlspecialchars function.

Quirks of Typolink

The typolink function already includes the use of the htmlspechialchars function, but not of the urlencodefunction. Also htmlspecialchars is inconsistantly applied. It is used for generating link tags but not for generating plain URL (property: returnLast). In the latter case you have to postprocess the generated URL with htmlspecialchars yourself, i.e. before you use it as a form action.

As we have seen above the processing by htmlspecialchars in the 2. step prerequests the previous urlending of keys and values. Because this is not done by the typolink function we urlencode keys and values for the additionalParams property as described above.

This are a few lines of version 0.0.20 of tx_lib_link:


if(!is_array($value)) { // TODO handle arrays
if($this->designatorString) {
$conf['additionalParams'] .= '&' . $this->designatorString . '['
. rawurlencode($key) . ']=' . rawurlencode($value);
} else {
$conf['additionalParams'] .= '&' .
rawurlencode($key) . '=' . rawurlencode($value);
}
}

As you observe in this version the square brackets are still not encoded to match RFC 1738.