normalize request urls for matching

Author: dannyvankooten

src/admin/class-admin.php

@@ -550,23 +550,7 @@ public function sanitize_settings($opts)
      */
     public function sanitize_url($url_string)
     {
-
-        // if empty, just return a slash
-        if (empty($url_string)) {
-            return '/';
-        }
-
-        // if string looks like an absolute URL, extract just the path
-        if (preg_match('/^((https|http)?\:\/\/)?(\w+\.)?\w+\.\w+\.*/i', $url_string)) {
-            // make sure URL has scheme prepended, to make parse_url() understand..
-            $url_string = 'https://' . str_replace([ 'http://', 'https://' ], '', $url_string);
-
-            // get just the path
-            $url_string = parse_url($url_string, PHP_URL_PATH);
-        }
-
-        // leading slash it
-        return $url_string;
+        return \boxzilla_normalize_relative_url($url_string);
     }
 
     /**

src/class-loader.php

@@ -108,7 +108,7 @@ protected function match_patterns($string, array $patterns, $contains = false)
      */
     protected function get_request_url()
     {
-        return rtrim($_SERVER['REQUEST_URI'], '/');
+        return \boxzilla_normalize_relative_url(isset($_SERVER['REQUEST_URI']) ? $_SERVER['REQUEST_URI'] : '/');
     }
 
     /**

src/functions.php

@@ -15,3 +15,68 @@ function boxzilla()
 
     return $instance;
 }
+
+/**
+ * Normalize a relative URL for Boxzilla rule storage and matching.
+ *
+ * @param string $url_string
+ *
+ * @return string
+ */
+function boxzilla_normalize_relative_url($url_string)
+{
+    $url_string = trim((string) $url_string);
+    if ($url_string === '') {
+        return '/';
+    }
+
+    if (strpos($url_string, '//') === 0) {
+        $url_string = 'https:' . $url_string;
+    } elseif (preg_match('/^[\w.-]+\.[a-z]{2,}(?:[\/?#]|$)/i', $url_string)) {
+        $url_string = 'https://' . $url_string;
+    }
+
+    $parts = wp_parse_url($url_string);
+    if (! is_array($parts)) {
+        return '/';
+    }
+
+    $path = isset($parts['path']) ? $parts['path'] : '/';
+    $path = '/' . ltrim((string) $path, '/');
+    $path = untrailingslashit($path);
+    if ($path === '') {
+        $path = '/';
+    }
+
+    if (empty($parts['query'])) {
+        return $path;
+    }
+
+    parse_str($parts['query'], $query_args);
+    if (empty($query_args)) {
+        return $path;
+    }
+
+    $tracking_keys = [
+        '_ga',
+        '_gl',
+        'dclid',
+        'fbclid',
+        'gclid',
+        'mc_cid',
+        'mc_eid',
+        'msclkid',
+    ];
+
+    foreach (array_keys($query_args) as $key) {
+        if (strpos($key, 'utm_') === 0 || in_array($key, $tracking_keys, true)) {
+            unset($query_args[ $key ]);
+        }
+    }
+
+    if (empty($query_args)) {
+        return $path;
+    }
+
+    return $path . '?' . build_query($query_args);
+}