return $url . (!endsWith($url, '/') ? '/' : '');
}
+/**
+ * Replace not whitelisted protocols by 'http://' from given URL.
+ *
+ * @param string $url URL to clean
+ * @param array $protocols List of allowed protocols (aside from http(s)).
+ *
+ * @return string URL with allowed protocol
+ */
+function whitelist_protocols($url, $protocols)
+{
+ if (startsWith($url, '?') || startsWith($url, '/')) {
+ return $url;
+ }
+ $protocols = array_merge(['http', 'https'], $protocols);
+ $protocol = preg_match('#^(\w+):/?/?#', $url, $match);
+ // Protocol not allowed: we remove it and replace it with http
+ if ($protocol === 1 && ! in_array($match[1], $protocols)) {
+ $url = str_replace($match[0], 'http://', $url);
+ } else if ($protocol !== 1) {
+ $url = 'http://' . $url;
+ }
+ return $url;
+}
+
/**
* URL representation and cleanup utilities
*
'action_type_map=',
'fb_',
'fb=',
+ 'PHPSESSID=',
// Scoop.it
'__scoop',
'utm_',
// ATInternet
- 'xtor='
+ 'xtor=',
+
+ // Other
+ 'campaign_'
);
private static $annoyingFragments = array(
return $this->toString();
}
+ /**
+ * Converts an URL with an International Domain Name host to a ASCII one.
+ * This requires PHP-intl. If it's not available, just returns this->cleanup().
+ *
+ * @return string converted cleaned up URL.
+ */
+ public function idnToAscii()
+ {
+ $out = $this->cleanup();
+ if (! function_exists('idn_to_ascii') || ! isset($this->parts['host'])) {
+ return $out;
+ }
+ $asciiHost = idn_to_ascii($this->parts['host']);
+ return str_replace($this->parts['host'], $asciiHost, $out);
+ }
+
/**
* Get URL scheme.
*
return $this->parts['scheme'];
}
+ /**
+ * Get URL host.
+ *
+ * @return string the URL host or false if none is provided.
+ */
+ public function getHost() {
+ if (empty($this->parts['host'])) {
+ return false;
+ }
+ return $this->parts['host'];
+ }
+
/**
* Test if the Url is an HTTP one.
*