<?php

/*
 * robots.txt
 * Note this isn't really a text file. The robots file specification doesn't allow for the use of entire domain names.
 * It only allows for the use of URL fragments, which don't include the host name. Hence, this script. This is parsed
 * as PHP via a directive in the default vhost config file. The goal is to prevent dl1-dl4, staging, and devs, from
 * being indexed, as well as parts of the web site that shouldn't be indexed, such as servlets and services.
 */

include_once "includes.php";

$productProperty = \productProperty::singleton();
$security = \security::singleton();
$wrapper = \wrapper::singleton();

// Set the output to "text/plain".
$productProperty->setContentType(\service::CONTENT_TYPE_TEXT_PLAIN);

$commonPatterns = "Disallow: /legacyservlets/
Disallow: /servlets/
Disallow: /s/
Disallow: /d/
Disallow: /modal-feedback
Disallow: /modal-login
Disallow: /modal-myaccount
Disallow: /modal-signup
Disallow: /modal-subscribe
Disallow: /organization-received
Disallow: /register
Disallow: /resetpw
Disallow: /verify
";

$clonePatterns = "Disallow: /data/our-data-directory
";

ob_start();

echo "User-Agent: *\n";

// We only allow indexing on PolicyMap proper and clones. Not on any other products (such as site licenses).
// We need to keep the first conditional as-is, because we can't have dl0-4 showing up in searches!
if($_SERVER['HTTP_HOST'] == "www.policymap.com") {

echo $commonPatterns;

} else if ($productProperty->getValue(\productProperty::VARIABLE_IS_CLONE)) {

echo $commonPatterns;
echo $clonePatterns;

} else {

?>
Disallow: /
<?php

}

$wrapper->content = ob_get_contents();
ob_end_clean();

?>