In working with updateviewcount.php in eZ Publish 4.0.2 I found it wouldn't parse my apache error logs correctly due to the logic used on line 161 and 162.

$urlArray = split( "/", $url );
$firstElement = $urlArray[0];

The above splits the requested URL and returns a blank value as the first array value if the resulting URL starts with a forward slash.

In the case of the log file I was parsing it was necessary to make the following change

$urlArray = split( "/", $url, 2 );
$firstElement = $urlArray[1];

This ensured that the first slash was removed and the resulting URL was passed to eZURLAliasML::fetchNodeIDByPath.

I think the logic is screwy and certainly isn't going to work in all cases, split shouldn't really be used to trim the leading slash and it probably isn't a required step at all. The existing code however seems flawed as it splits the URL on forward slashes and only passes the first value onto eZURLAliasML::fetchNodeIDByPath. This won't work as it will only register visits to top level nodes and not their children as these paths are stripped and not passed through to the function.